The Ground-Floor Myth

Engineers often tell us they want to join a company at the ground floor. The instinct makes sense. Join early enough and your equity could be meaningful. You get to shape the product, influence the culture, and say you were there before the company had a proper logo.

But early and high-upside are not always the same thing. The ground floor comes with the risk that the building never gets built, the product may not work, and the market only exists in the fundraising deck.

There’s another compelling moment: when a company has real traction, customers, and evidence that it’s working—but still has an enormous amount left to build.

A billion-dollar valuation can sound like the party is over. Sometimes it means the product finally works and the party got a liquor license.

After the Signal Is Clear

On a recent podcast, Founders Fund’s Brian Singerman described a new perspective on venture investing: the hardest part is not only finding the winners early. It is continuing to recognize the winner after it stops looking early – Founder’s fund put roughly one-third of a fund into Palantir at a $1 billion valuation.

The same principle applies to choosing where to work.

An engineer considering SpaceX or Anthropic after the product and market were credible might have worried they missed the ground floor. But equity does not need to start cheap to become valuable. FYSK featured Anthropic in February 2023, when it was worth roughly $5 billion. It was not exactly hiding in a garage. Today, Anthropic is valued at $965 billion. Holding ownership constant, $100,000 of equity maps to roughly $19 million today. The ground floor was gone. Apparently, almost none of the upside was.

Serval and Owner are far from being companies with three people and a deck. They have products, customers, and incredible momentum. But the larger markets they’re pursuing are far from finished. It’s time to change your mindset from “What has already been done” to “How much is there left to do.”

General Matter

General Matter may be the clearest example.

Scott Nolan founded the company to rebuild America’s domestic nuclear-fuel capabilities. It is focused on uranium enrichment and advanced nuclear fuel production, one of the most important bottlenecks facing the next generation of nuclear energy. The company has secured a $900 million Department of Energy task order and raised $50 million in a round led by Founders Fund.

A meaningful amount has already been proven.

But the hard work is not over. General Matter still has to build complex physical infrastructure, develop and operate enrichment technology, navigate a heavily regulated industry, and help restore a critical part of the American industrial base.

That is not getting in after the adventure. That is the adventure.

The best startup jobs are not always at companies nobody has heard of yet.

Sometimes they’re at companies everyone can see are working—but few people understand how far they still have to run.

Apply here if you’re interested in dinner with General matter or future FYSK dinners.

A company’s website can tell you what it builds. It can’t tell you that a hardware engineer liked Flow enough to leave his job for a company that would let him use it. Or that the right person to reinvent construction may be an aerospace engineer. Or what keeps a healthcare team working late when the work stops being theoretical.

That’s why we started hosting dinners.

Unlimited Industries

Our first dinner of the Summer was at Unlimited’s office with Alex Modon and Tara Viswanathan.

Their manifesto is about making the physical world move faster. Their office felt like proof they meant it. Team members arrived from calls, disappeared to run models, and returned to the table—genuinely welcoming, but visibly eager to get back to building.

In between moments like an attendee realizing he and Tara were from the same hometown, Alex and Tara shared Unlimited’s progress. They’re already compressing processes from months into weeks, but were candid about the thing AI cannot shortcut: credibility. Until you have completed enough projects, knowing the best way to build something is not the same as being trusted to say so.

Flow Engineering

Next was Pari Singh of Flow Engineering.

Pari told two wild stories you will not find about Flow online. In one, a hardware engineer changed jobs because his company would not let him use Flow. In another, Flow was rejected after a long product evaluation—only for the engineers to push leadership to reopen the decision. Flow came back and won.

The company names stay at the table.

Pari described Flow as the kind of family that argues, works through the issue, then hugs it out. That tracked: intense but warm, and caring enough to disagree properly.

Healthier

Then we had dinner with Parth Bhakta of Healthier. They’re building an AI reasoning platform for oncology—bringing together fragmented medical records, imaging, genomics, research, and clinical guidelines into recommendations patients and doctors can actually evaluate.

The views of the Bay Bridge were incredible, the dinner was delicious, but the personal stories are what stuck.

A friend of a founding team member once called them after his father was diagnosed with cancer. He had great doctors. He still needed help. It’s the kind of call that makes staying late feel different.

Parth made the point that hard things are interesting precisely because they are hard. The team seems to attract people who are empathetic without needing to make a production of it. They are not working toward some distant hypothetical impact.

Someone needs the product now.

You can learn plenty about a company online.

You cannot search how the team argues, what makes them stay late, or whether the room feels more interesting two hours in.

For that, you have to come to dinner.

Apply here to join our dinner series if you would rather catch the vibes in person than read our reconstruction.

Play FYSK Builder deck

Make eight fast choices, draw your engineering character card, and unlock the FYSK companies best matched to the way you build. Like Myers-Briggs (but way shorter!) the game takes about two minutes and reveals where you thrive, how you work, and the particular form of technical wizardry you bring to a team.

Your result comes with its own:

BUILD ENGINE
How you turn ideas into working things.

BATTLECRY
The phrase your coworkers have heard 4,000 times.

INSTINCT
What your brain does before the meeting has technically started.

HABITAT
The environment where your powers become most powerful.

Resist the urge to add this card to your LinkedIn certifications. Actually never mind…do it. LinkedIn could use some more adorable, colorful, magical animal cards!

Once you’ve taken the quiz, we find your people

This is where the personality quiz develops a useful side quest. Your builder type, preferred industries, and ideal company culture are matched against every company that has presented at Founders You Should Know. Years of exceptional startups curated for you by one highly opinionated matchmaking machine.

Unlock bespoke connections

When a match looks promising, ask FYSK for a private introduction. We’ll approach the founder, brag about you appropriately, and make the connection when there’s mutual interest. Learning that you are a highly evolved Infrastructure Beast is fun. But learning that your ideal team is already hiring is better.

Get in on the magic

Discover your engineering personality. Meet the founders you should know.

Play FYSK Builder deck

This only takes two minutes — unless you attempt to reverse-engineer the scoring system—which would also tell us quite a lot about you.

Bonus content: notes from my very first vibe-coded project!

This was my first time vibecoding and deploying an actual product. I know almost nothing about engineering, which made the process feel equal parts magical and frustrating. It reminded me of baking—an activity I do not love. There’s room to be creative, but getting anything into production requires following a long list of very exact instructions. Miss one step, and the whole thing refuses to rise. Except instead of a sunken cake, you get a 404.

A few things I learned:

The artwork needed an actual design system.
The first versions were all over the place, so I had to define a proper design file and keep refining it until the cards finally looked like they belonged in the same universe.

Ugh, ugly v1, or v25…

“Every FYSK company” turned out to require some human judgment.
The original database included companies that had been acquired or shut down. I asked Codex to scour public websites for a first pass, then manually reviewed and removed the ones that should no longer appear in the matches.

AI loves putting too much text on the cards.
Early versions were crowded, distracting, and full of lines no human would actually say. A large portion of the work was deleting things, simplifying the layout, and rewriting copy until it sounded like a person instead of a very enthusiastic machine.

What does the text in the little box windows even mean?

The badges took forever.
Getting them to feel culturally relevant, funny, distinctive, and actually connected to each engineering personality required many, many iterations. Apparently “make it cooler” is not a complete creative brief, despite my repeated attempts to make it one.

Deploying the game was the hardest part.
Because I had no idea what I was doing, I initially asked for the simplest possible setup and was guided toward Netlify with a private Airtable handling the backend storage.

Then the human in the loop—my husband, a seasoned vibecoder—suggested GitHub, Vercel, and Supabase instead.

What followed was a painstaking sequence of instructions, failed steps, screenshots uploaded to Codex, revised instructions, more screenshots, more revised instructions, and several moments where every button on the internet appeared to be in the wrong place.

Eventually, it worked (mostly!)

There are still bugs. Please report them (seriously!) Finding a bug is now part of the game.

Not the right Detail! Ironically we could use Detail to find the bugs…

The connections we made mattered, but both candidates had plenty of other options and closing them was entirely in the founder’s hands. We can lay up an introduction, but like a first date, what happens next depends on the people in the room.

The best founders understand that. They treat the interview process as a chance to build trust, create momentum, and show exceptional candidates why this is the company – and team – they should bet their career on.

How great founders win great people:

1. Speed as a weapon
   The fastest hiring outcomes we’ve seen are highly intentional. One founder closed an exceptional engineer within 48 hours of meeting him at a FYSK showcase. Another hired two engineers the week after we hosted a dinner for the company. And the founder we mentioned above closed both candidates within a week of our introductions.
   
   These founders were building companies, shipping product, talking to customers, and managing a hundred other priorities, but they treated promising candidates with urgency. They followed up immediately, created momentum, and made every next step clear.
   
   We know that is easier said than done. Recruiting is one of many things competing for a founder’s attention, and it is easy for a follow-up to slip. But exceptional candidates move quickly, and the founders who move with them are often the ones who win.

   
   A strong candidate should never be wondering when they’ll hear from you. If a candidate has to nudge you, the process has already lost momentum. Make it a goal to keep things moving whenever the ball is in your court.
   

2. Sell the dream, not the job description.
   Great founders do more than explain what the company is building today. They make the future feel tangible.
   
   They do not lead with, “We’re looking for a Python backend engineer.” Almost no one wakes up identifying as a Python backend engineer. They lead with the problem, the stakes, and the future someone could help create.
   
   Compare that with: “Energy is the foundation of every AI system. We’re working on uranium enrichment to unlock abundant, clean power—and if we succeed, the future of AI runs on infrastructure we helped make possible.” They might even add: “By the way, our backend is built in Python. Even if you do not know it yet, I’m sure you can figure it out.”
   
   The best engineers are looking for the opportunity to help build a generational company. They want to understand what this could become, why it matters, and why this team has a real chance of making it happen.

   
   3. Don’t make candidates play detective.
   Strong candidates will ask about revenue, growth, compensation, competition, risks, and the challenges ahead. Answer those questions clearly. Engineers notice when founders are evasive. We recently heard from a candidate who asked about compensation and was told the company was “competitive,” but was not given a range. When they pressed, the numbers shared felt surprisingly low relative to the company’s massive recent funding round. Their takeaway was harsh: the process ranked “bottom decile” on both compensation and transparency. They still agreed to continue the conversation but the founder had already created doubt that did not need to be there.
   
   If you are asking someone to devote years of their life to building alongside you, it is hard to earn that commitment while gatekeeping the information they need to evaluate the opportunity.

   
   Transparency during the interview process signals how you will communicate once they join and begins establishing the culture of the company before their first day.
   

3. Let your team do the selling.
   Candidates are evaluating the people as much as the company. Give them meaningful time with the strongest members of your team, especially the people they would work with most closely.
   
   We recently heard this from an engineer after meeting one of the FYSK-featured companies: “The idea was not an immediate ‘wow,’ but it also didn’t strike me as a bad direction. For me, it came down to the passion, charisma, and caliber/pedigree of the team. Once I met more people on the team, I was so much more excited about working there.”
   
   Exceptional engineers want to collaborate with exceptional peers. They want to know who they’ll be solving hard problems with, who they’ll learn from, and whether this is a group that will push them to do the best work of their career. People want to join the coolest, most interesting, most brilliant club they can get into.
   

4. Don’t lose on the offer.
   The strongest engineers are often considering several opportunities at once—and they rarely evaluate an offer alone. They share it with friends, former colleagues, founders, and other people who have seen dozens of competing packages.
   

   One engineer recently told us that while he was assessing his offer, a founder friend was blunt: the comp package he received was far too low, he was worth significantly more, and the alternatives were better. Because that friend had seen many comparable offers and knew him well, the feedback carried real weight.
   
   A low-ball offer, especially one with weak equity, can quickly remove you from consideration. Research the market and compensate early employees appropriately for the risk they are taking. Raising significant capital from a top-tier investor does not automatically make the opportunity feel derisked to a candidate.
   

   Early employees only get to make one bet with their time. Your offer should reflect the value of that bet.

And lastly…remember every candidate talks.

An interview process leaves an impression far beyond the candidate in the room. Treat every candidate as someone who will share the experience with 30 other exceptionally talented people (because they often will!). Engineers talk to their friends, roommates, partners, former colleagues, and to us at FYSK. When founders run a thoughtful, respectful process, word travels, and it becomes easier to attract even more exceptional people.

Because we’re often the connector, we care deeply about what happens after the introduction. The founders who consistently make candidates feel valued are the ones we’re most excited to keep introducing to great talent. Run a great process, and the flywheel starts to work in your favor.

Thank you to Camille Ricketts for chatting with me about this topic! For more of Camille’s musings on talent, company-building, and the occasional extremely important TV or movie recommendation, check out their newsletter, Early Days.

LinkedIn is fine, but personal websites are where the weird good stuff lives. AI has made building easier, which means the bar has moved. The most interesting people now show receipts: projects, writing, design taste, weird obsessions, half-finished tools, thoughtful essays, and GitHub rabbit holes.

That’s part of why personal websites are having a moment in the FYSK community. LinkedIn still has its place. It tells you someone’s title, company, school, and maybe whether they recently discovered the phrase “AI-native.” Helpful and fine. A little beige.

The personal websites we’ve seen show how the creator thinks. The best ones feel less like resumes and more like personal trophy cases: things built, people built with, ideas chased, side quests pursued, and tiny clues that make it clear there’s a human behind the keyboard.

Here are a few we loved:

Bryce Sandlund

brycesandlund.com

Bryce’s site proves he can’t be reduced to a LinkedIn headline. There are research papers, software projects, competitive-programming history, opinions on academia, AI writing, and old projects. There’s also beer brewing, music, and outdoor adventures.

Almost none of this would fit on a normal resume. The research is serious and substantial, but the site never flattens him into “person with publications.” You can follow links into reinforcement learning and AI, explore independently built products, see the competitive-programming community he’s participated in and supported, and then casually discover that he also makes tikis (Bryce if you’re reading this, I want one). The site gives you a much fuller picture of Bryce: not just what he has done, but what he notices, what he builds, and what corners of the internet his curiosity has colonized.

Harsha Karanth

harshakaranth.com

Our favorite part of Harsha’s site is the project section. The work is concrete, clickable, and pleasantly all over the place. There’s Notefolio, an investment journaling platform linked to the App Store. Cardiac AR, which points to research on collaborative mobile AR for cardiovascular surgical planning. There’s also an AI-powered employee onboarding platform with a video demo, an SEC 8-K stock predictor, and a mobile version of Egyptian Rat Screw – which we can only assume is safer for hands than the card-slapping original.

By the time you reach the line “reaching for the stars, one line of code at a time,” the pun feels earned. And for the record, we are pro-pun: Harsha makes a strong case that he is, in fact, a star.

Cooper Shea

coopershea.com

Cooper’s site makes you feel like you just dropped into someone’s workshop and realize they have five tabs open in their brain. He’s obsessed with questions around computational design, manufacturing, materials, AI, human cognition, and the messy process of turning ideas into physical things. Then you hit the projects tab and the builder brain fully reveals itself.

There’s a cleft palate orthodontic device for infants with Pierre Robin sequence, built through a Blender, nTopology, and Python pipeline that cut a manual process from 14 hours to 2. Right next to it is a Monstera leaf bottle opener, VR-sculpted in Gravity Sketch and post-processed in Fusion 360. The range is impressive. You get medical devices, aquaculture robotics, hydrogen systems, microneedles, ocean plastics recovery, and houseplant-shaped bottle opener. It’s a very specific flavor of builder energy: deeply technical, physically grounded, and just weird enough to be memorable.

The CV is tucked away at the end. After clicking through projects, theses, writing, and the homepage quiz — which took us way too long, no further questions — you eventually learn Cooper has a Stanford M.S. in Mechanical Engineering focused on Biomechanics and Thermodynamics. By then, the credential lands differently. Like “oh, that explains the houseplant bottle opener pipeline.”

Madhuri Bhavana

madhuribhavana.com

Madhuri’s site is a great example of a portfolio that shows not just what the final product looks like, but how the thinking happened. Her work is organized around real product problems: redesigning a complex virtual-machine creation flow at Nutanix, designing Amazon Easy for shoppers in emerging markets, and conducting research for Google’s AdMob platform.

The case studies do what good case studies should do and what many case studies heroically avoid doing: they show the actual judgment. They lay out the problem, users, research, insights, tradeoffs, and concept development alongside the final experience. You can see the difference between someone who can make a clean interface and someone who can make complicated systems more understandable, inclusive, and useful. That’s the real signal.

The best personal websites do something resumes and LinkedIn rarely do. They remind us there is a person behind the page. They show the projects behind the titles, the taste behind the work, the curiosity behind the credentials, and the side quests that make someone memorable.

That’s part of what makes the FYSK community special. Yes, the talent is exceptional. But the real magic is the density of curiosity, generosity, ambition, taste, and “you really built that?” energy in one place.

Thanks again for being a part of FYSK!

Have a personal website you’re proud of? Share it when you apply to the FYSK community!

Hungry to meet founders and can’t wait until the next showcase in September? Join FYSK and other jobseekers for an intimate dinner with six of our favorite founders. We chose these founders because their companies have incredible momentum: Over 130% YoY revenue growth, hundreds of millions in government support, backed by a16z, Sequoia, Founders Fund, and Nat Friedman – yet they’re still early enough for new hires to shape the direction of the company. Each founder is actively hiring and looking to connect with candidates from the FYSK community!

These dinners are intentionally small. They’re an exclusive opportunity to speak directly with the founder, a few early team members, and ~10-15 other engineers also interested in attending. You’ll have a chance to personally feel out company culture, understand the technical problems the team is working on, and maybe even make a new friend.

Here’s what attendees from our most recent dinner series had to say:

This Summer’s Menu

These teams are growing quickly. Now is your chance to join and have real ownership and an outsized impact.

Tara Viswanathan, Co-Founder of Unlimited

📅 July 16 | San Francisco

Unlimited is an AI-native engineering and construction company that leverages software and artificial intelligence to improve how factories, data centers, energy infrastructure, and other critical projects are designed and built. Tara and the team’s vision is to transform the delivery of large-scale infrastructure by combining AI, software, and engineering expertise.

Unlimited recently raised a $12 million seed round co-led by Andreessen Horowitz and CIV and is already delivering measurable results on industrial projects, reducing pre-construction engineering timelines from six months to a few weeks and identifying design optimizations that cut costs by over 50%.

Pari Singh, Founder & CEO of Flow Engineering

📅 July 21 | San Francisco

Pari is building Flow Engineering to accelerate the development of the world’s most important physical technologies. Flow helps engineering teams design, test, verify, and iterate on complex hardware systems dramatically faster, enabling companies building rockets, autonomous systems, energy technologies, and advanced manufacturing products to operate with software-like speed. The cost of software creation collapsed once AI could write code, and Flow’s bet is hardware engineering will undergo a similar transition.

Flow’s software is already deployed in many of the top next-gen hardware companies. Rivian evaluated 30 different tools and ultimately chose Flow. They started using Flow with only 40 users and grew to 1,500 in four months.

Parth Bhakta, Founder & CEO of Healthier

📅 July 23 | San Francisco

Parth is building Healthier, a trusted AI reasoning platform for high-stakes medicine, beginning with oncology. The company’s technology synthesizes fragmented medical records, imaging, genomics, clinical guidelines, research literature, and trial protocols into source-grounded recommendations that both patients and clinicians can understand and evaluate. Healthier’s products span patient-facing care navigation, AI-powered clinical decision support, and tumor board workflows designed to help providers make more informed decisions in complex cases.

We’re excited by the team and their early commercial traction. The team consists of advisors and operators from Stanford, UCSF, Mass General, Dana-Farber, UPenn, and Foundation Medicine. They recently hired a Chief Medical Officer who previously worked at OpenEvidence and City of Hope. It’s still early, but they already have pilots and partnerships with major oncology networks and institutions and several top cancer centers. The company recently closed a seed round led by a16z.

Rune Kvist, Co-Founder of AIUC

📅 July 30 | San Francisco

Rune is building AIUC, a company creating the trust infrastructure for AI. As enterprises deploy increasingly capable AI systems and autonomous agents, AIUC is developing standards, certification frameworks, and insurance products designed to help organizations evaluate and manage AI-related risk. The company is pioneering a new category of infrastructure intended to make advanced AI deployment safer, more transparent, and more accountable.

AIUC combines a strong founding team (experience spanning Anthropic, McKinsey, METR) with exceptional early traction. The company already launched the AIUC-1 standard, raised a $15M seed round, and is working with some of the most prominent AI companies in the world to define how enterprises certify, insure, and adopt AI agents.

Adam Guild, Co-Founder & CEO of Owner

📅 Date: tba | San Francisco

Adam founded Owner after helping save his mother’s small business and realizing how difficult it had become for independent businesses to compete online. Owner helps restaurants grow direct revenue, strengthen customer relationships, and compete against large chains using AI-powered tools and software infrastructure. What began as a mission to empower local business owners has evolved into a platform serving independent restaurants across the country.

Owner now serves more than 10,000 restaurants across the United States and recently raised a $120 million Series C at a $1 billion valuation.

Scott Nolan, Founder & CEO of General Matter

📅 Date: tba | Los Angeles

Scott founded General Matter to rebuild America’s domestic nuclear fuel capabilities. The company is focused on uranium enrichment and advanced nuclear fuel production, addressing one of the most critical bottlenecks facing the future of nuclear energy. As countries race to deploy next-generation reactors and strengthen energy security, General Matter is working to ensure the United States has the industrial capacity necessary to support a new era of nuclear power.

General Matter secured a $900 million Department of Energy award to help rebuild domestic HALEU enrichment capacity, and they recently raised a $50 million round led by Founders Fund, with Peter Thiel joining the board.

Interested in attending? Apply here and let us know which dinners you’d like to join!

Happy SpaceX IPO day! A multi-trillion valuation is a remarkable achievement by any measure, but our hot take is that among the most valuable legacies SpaceX leaves in its plume is its alumni founder network. SpaceX has become the launchpad for a new wave of founders building some of the wildest companies in the world.

Looking back, you can identify companies that shaped entire generations of entrepreneurship. Fairchild Semiconductor produced the founders who built Intel, AMD, and dozens of companies that became the backbone of Silicon Valley. PayPal (notably also founded by Elon!) became the training ground for founders who would go on to build Tesla, LinkedIn, Palantir, Yelp, YouTube, and countless others. These companies became environments that further shaped already talented people, helping them develop a new understanding of what’s possible. SpaceX is one of them.

Building rockets forces people to think across software, manufacturing, operations, supply chains, and energy systems at once. Few organizations force people to operate across so many disciplines simultaneously. But it’s also cultural. SpaceX had a reputation for assigning enormous responsibility to young people and expecting them to solve problems that most organizations would consider impossible.

The result is founders willing to tackle incredibly complex problems. Here are a few SpaceX alum founders that we’ve featured at our FYSK showcases and a couple others who made the news recently!

General Matter — Scott Nolan

As electricity demand surges and energy security becomes increasingly important, one of the biggest bottlenecks in nuclear power is fuel, not reactors. General Matter is focused on domestic uranium enrichment, tackling a deeply technical and highly regulated challenge that sits at the foundation of America’s energy future.

We featured General Matter at the most recent showcase. Scott is the type of founder we’re proud to feature: ambitious, deeply technical, and focused on a generational problem.

Base Power — Justin Lopas

The electric grid is becoming more complex by the year as AI data centers, electrification, and extreme weather put new pressure on aging infrastructure. Justin Lopas and the team at Base Power are building distributed energy systems designed for a grid that will be more resilient, intelligent, and decentralized than the one we have today.

We were excited to have Justin present Base Power at FYSK last October; building the infrastructure needed for an increasingly electrified world is both technically difficult and enormously important.

K2 Space — Neel Kunjur

For decades, satellites were treated as custom-built aerospace projects. K2 Space believes the future requires something different.

The company is building large, high-power satellites designed to support the growing demand for communications, defense, and national security capabilities in orbit. As space becomes critical infrastructure, K2 is helping build the backbone.

That thesis was validated this week when K2 was selected as a satellite provider for the U.S. Space Force’s Protected Tactical SATCOM-Global program.

Starcloud — Adi Oltean

Every founder ecosystem produces a few ideas that sound ridiculous at first. Starcloud is one of them.

Adi Oltean is building data centers in space. It’s a concept that feels pulled from science fiction until you remember that reusable rockets once did too. And Starcloud isn’t just talking about the future – the company launched a satellite carrying an Nvidia H100 GPU into space last November.

Starcloud recently raised a $170 million Series A at a $1.1 billion valuation, reaching unicorn status less than two years after graduating from Y Combinator.

A well-deserved congratulations to all SpaceX alumni for such a remarkable milestone! This is just a small sampling of a much larger set of ex-SpaceX founders to explore. We’re excited by the number of founders taking lessons learned from SpaceX to new problems and we look forward to seeing many more.

If you know (or are!) a SpaceX alum founder building in energy, defense, healthcare, or are focused on another foundational challenge, we’d love to get to know you!

We recently received this feedback from a long-time community member. She left her previous role to join a startup she discovered through the FYSK YouTube channel. Community members often ask us how to find the companies we’ve previously featured, and the FYSK YouTube channel is the perfect resource.

Go to the channel and watch a few founder presentations from a showcase you missed. If you find a mission or problem that resonates, don’t hesitate to reach out to us! Reference the video and explain what made you interested. Founders at this stage of their company care deeply about who joins them, and a genuine, specific message goes a long way.

If you’re actively looking for a job this summer and having FOMO because the next showcase isn’t until September, look no further.

Your summer watchlist here!!

The whole point of FYSK has always been to get the right jobseekers in front of the top founders at the right time. The showcase is often where FYSK members first discover companies they go on to work for, but the opportunity doesn’t disappear once the event ends. Since founders only present once, it’s easy to feel like you’ve missed your chance if you couldn’t attend live. Fortunately, that’s not the case — every founder presentation remains available on our YouTube channel. It’s a growing library of exceptional founders and startups you can explore whenever you're actively thinking about your next role.

You can find presentations from the founders of Cursor, Anthropic, Parallel AI, Exa, and dozens more. Those videos are still generating conversations today. Some of the people who watched those videos and reached out are now engineers, PMs, and designers at those companies.

Looking for your next role?

• Watch a few founder presentations.

• Find a company that genuinely excites you.

• Send us a note explaining why.

If there’s a strong fit, we’ll help make an introduction.

We look forward to hearing from you!!

We’ve been hearing people say that the big AI labs are killing startup opportunities – adding new capabilities into their models so fast that startups can’t compete, and hoovering up all the talent that would otherwise be excited to join them.

We take the exact opposite view. We think there’s never been a better time to build something impactful at a startup. If you’ve been coming to FYSK for a while, you know we featured Anthropic back in 2023, when it was still under 100 people and a Series B company. Joining Anthropic at that stage turned out to be an incredible startup opportunity, and anyone who did is probably feeling very good about that decision today. But what’s happened since is even more important. The foundation models have become a utility like water and electricity — widely accessible with just an API call. They’re tools that everyone can now build with. And that means a whole new categories of ambitious startups have become possible.

The founders we featured last week are tackling some of humanity’s biggest and hardest problems: how to create abundant clean energy, how to foster deeper human connection, how to build at the speed of thought, and how to design the next generation of rockets and spaceships.

Talented engineers are finding fresh opportunities beyond the frontier labs at these startups. On Wednesday, we celebrated two community members who found their next chapter through FYSK. Sara Pohland joined Voio after attending a dinner we hosted with the founder, and Aadarsh Pratik joined Campfire after meeting them at the October showcase. Both received their FYSK hoodies on stage (and you can too if you’ve gotten a job through FYSK)!

If you’d like to be connected to any of the founders below, send us an email blurb with context and we’ll reach out to the founder to facilitate a connection.

Flow Engineering

Flow Engineering helps hardware teams design, test, and iterate complex systems at software speed.

Founder: Pari Singh

Open Roles: Account Executives, Software Engineers (Frontend, AI, Infrastructure, Full stack, Forward Deployed)

Keycard

Keycard builds identity and access infrastructure for AI agents, helping developers securely connect agents, people, and systems.

Founder: Ian Livingstone

Open Roles: SDK engineer, AI/Agent engineer, Systems engineer, Frontend engineer

Eigen

Eigen is building “the world’s mutual friend” — an AI companion designed to help people belong, grow together, and connect more deeply with one another.

Founder: Paul Scherer

Open Roles: Engineering, Operations

Serval

Serval builds AI agents for IT service management.

Founder: Jake Stauch

Open Roles: Software Engineers, Forward Deployed Engineers, Deployment Strategists, Designers, Account Executives, other

Tasklet

Tasklet is the agent OS for knowledge work. Scale your company without adding headcount with AI agents that do the work for you.

Founder: Andrew Lee

Open Roles: Agent engineers, Design engineers, Product engineers, Product Manager, GTM

General Matter

General Matter is fixing the domestic fuel bottleneck in nuclear energy through uranium enrichment infrastructure.

Founder: Scott Nolan

Open Roles: Hardware Engineering, Software Engineering

Sail Research

Sail Research provides the infrastructure for abundant intelligence.

Founder: Neil Movva

Open Roles: Distributed Systems, GPU Optimizers, Backend Generalists, Rust Lovers

Specter

Specter provides real-time physical intelligence across critical infrastructure and industrial environments through wireless long range sensing and general AI.

Founder: Xerxes Libsch

Open Roles: Software Engineering, Hardware Engineering, Manufacturing

OpenRouter

OpenRouter is a marketplace and unified API for accessing, comparing, and routing across hundreds of AI models.

Founder: Alex Atallah

Open Roles: Software Engineer (Platform, Product, Dev Products, Full Stack), Data Engineer, Security Engineer, Growth Engineer, Developer Relations Lead

Conduit

Conduit is building mind-reading AI for high-bandwidth brain-to-computer communication through a non-invasive thought-to-text system.

Founder: Rio Popper

Open Roles: ML Research, ML Ops, ML Infra, Data Engineers

Thank you again to everyone who attended and to Rippling and Digital Ocean for their continued generosity in supporting our community!

Couldn’t make this showcase? Feeling the FYSK FOMO? Apply now to our community and subscribe to our newsletter to get your invite to private founder dinners and our next show!

AI is rewriting how humans interact with software, work, and even each other. Conduit is building an interface for direct brain-to-computer communication. Tasklet is replacing brittle SaaS workflows with autonomous agents. OpenRouter is building a routing layer for the exploding universe of AI models. And Eigen is building AI to strengthen human relationships instead of replacing them. Together, they are reimagining a future where AI more seamlessly integrates with everyday life.

Rio Popper, founder of Conduit Intelligence

For all the advances AI has made over the past few years, the way we interact with computers has hardly changed. We type on keyboards and tap touchscreens. At the leading edge, there’s speech-to-text and autocomplete, but none of it can match the speed of thought.

Rio Popper, founder of Conduit, believes that bottleneck is limiting human potential. After becoming blind at age five, she learned to navigate technology differently from everyone else. She built a motion sensing interface to join the swim team, and went on to create an unusually high-speed workflow for coding. With this tool, she found herself coding faster than many sighted engineers. These experiences made her realize how outdated modern interfaces really are.

Conduit grew from this realization – it’s a radically different interface where software operates at the speed of thought. The company is developing non-invasive neural interfaces that allow AI systems to interpret neural signals directly and translate them into action. At the foundation of the project is over 10k hours of data from thousands of unique individuals, comprising the largest neuro-language dataset in the world. Early prototypes can already infer rough semantic meaning directly from neural signals.

Meet Rio at the showcase to learn more about creating neural interfaces for the AI era.

Andrew Lee, Founder of Tasklet

AI chatbots can answer questions and generate text, but unfortunately they don’t actually do your work for you. You’re still manually tipping every other domino in a disjointed Rube Goldberg machine of SaaS apps. Today, most people spend hours every day copying information across apps, updating CRMs, triaging emails, and stitching together workflows that should be automated. Traditional automation tools have cut some of this, but they require brittle flowcharts and constant maintenance over new edge cases.

Tasklet is the alternative. Andrew Lee built it as an AI agent that connects every tool you use and actually do the work. Tasklet operates as a cloud-native AI agent that can take real action across the tools people already use. It connects through thousands of integrations, APIs, MCP servers, or even a full browser. It can generate interfaces, write and execute code, process data, and run workflows continuously in the background. The best part is how simple the interface is. It feels like typing into a basic LLM chat box, but behind the scenes it’s actually running workflows and taking actions. Knowledge work is shifting away from humans operating software to agents doing it for them, and Tasklet is emerging as the operating system to coordinate it.

If you’re interested in building the operating system for the future of work, come meet Andrew at the showcase!

Alex Atallah, founder of OpenRouter

In the words of Alex Atallah, there’s been “a Cambrian explosion of models,” and with the boom comes a constantly shifting optimization problem. There is a new release every week from OpenAI, Anthropic, Meta, Google, DeepSeek, Mistral, and open-source variants, each with different pricing, latency, reliability, and infrastructure requirements. There is no ‘best’ model for every task and regular updates make choosing one even harder. A model that’s cheapest today may be obsolete next month. One provider might be fastest for simple queries but comparatively fail on reasoning. Another may offer better privacy but higher latency. As inference costs become one of the largest operating expenses for AI-native companies, developers are spending increasing amounts of time building routing logic, benchmarking providers, and continuously reevaluating which models to use.

OpenRouter is solving the optimization problem by becoming the universal routing layer for AI. It eliminates the need for developers to hardcode themselves to a single model provider, giving developers a single API endpoint connected to hundreds of models and dozens of infrastructure providers. It intelligently routes requests based on price, speed, reliability, or performance requirements, connecting you with the right model, for the right query, at the right time. OpenRouter aims to become a sort of AWS for inference traffic – a neutral marketplace between applications and the universe of base models.

Meet Alex at the showcase for a chance to build the universal API for the AI era!

Paul Scherer, Founder of Eigen

Social media apps designed to bring people together now feel flooded with algorithmic recommendations and push engagement over genuine connection. AI companion apps and personal chatbots threaten to accelerate this fragmentation further, offering a substitute for friendship rather than strengthening your real ones. Most companies are racing to build “your AI best friend,” without thinking about the collective experiences that make us human.

Paul Scherer at Eigen is trying to build the alternative. For him, the goal is to create shared experiences that bring people together rather than silo them. He’s creating “the world’s mutual friend.” Imagine you’ve just moved into an apartment and are looking for furniture. Your ‘mutual friend’ gives you more than generic aesthetic advice. It’ll connect you with your friend that it knows is also going furniture shopping today. Looking for ways to publicize a showcase? Instead of GPT-esque bullet points on ‘when to start promotion’ or ‘what graphics to use,’ your mutual friend reaches out to your founder friend who’s already hosted ten and comes back with a real opinion. Eigen imagines technology that nudges people toward shared discovery, introduces friends-of-friends, encourages collaborative experiences, and helps people feel more connected to the communities around them. It’s an AI that understands not just information, but your relationships, trust, and taste. In a world increasingly shaped by algorithmic isolation, Eigen is betting there’s enormous demand for technology that helps people belong and grow together instead.

If building technology that makes people feel more connected sounds like your vibe, come meet Paul at the showcase!

Apply to attend our next SF Startup Showcase and meet these founders in person.

Energy systems, industrial operations, and hardware development were never designed for the scale and complexity of the AI era. Critical infrastructure is still monitored through fragmented legacy systems with limited real-time visibility. Hardware engineering continues to rely on manual workflows, and as electricity demand surges from AI infrastructure, the U.S. faces gaps in its energy supply chain. These three founders are building solutions to these foundational problems by applying modern software, AI, and industrial technology to the core physical systems that power the economy.

Xerxes Libsch, Founder of Specter

Much of the world’s critical infrastructure is monitored by a few cameras and maybe a security guard intermittently checking a monitor. There are power plants, industrial sites, oil rigs, nuclear plants, and data centers that still rely on legacy monitoring systems, leaving operators with incomplete information, delayed alerts, or costly manual monitoring. The result is intrusions, safety incidents, or equipment failures that are detected too late, if at all. Software has transformed digital systems, but the physical world remains largely unobserved.

Xerxes is taking his experience from Uber and Anduril to solve this problem with Specter, a real-time perception system for the physical world. By deploying a network of low-cost, rapidly installable sensors connected through a wireless mesh, Specter gives companies continuous visibility across their entire physical footprint. Their platform acts as a software-defined control plane, turning raw sensor data into real-time intelligence that can detect intrusions, monitor operations, and enable automated responses.

Picture a system that partners with AI to detect everything from missing safety gear on a construction site to a building fire or a fight in a crowded stadium. The result is a shift from reactive, fragmented monitoring of physical systems to software-level observability and control.

If you’re curious to learn more about bringing real-time perception and control to the physical world, come meet Xerxes at the next Founders You Should Know showcase!

Pari Singh, Founder of Flow Engineering

The world around us relies on hardware engineering, from cars and nuclear reactors to the device you’re reading this on. Yet the system it was built with is still stuck in the dark ages. Walk into one of the best hardware engineering organizations in the world (think Space X, Tesla) and you will find teams of brilliant engineers manually moving between Excel spreadsheets, CAD tools, simulation environments, PDFs, and compliance documents. It’s extremely manual. While software development has undergone a dramatic transformation through AI-assisted coding and autonomous agents, physical engineering has largely been left behind.

As a mechanical engineer himself, this problem drew Pari Singh to develop Flow Engineering. With Flow, we are entering a world where intelligent agents will not just write software, but also design mechanical systems, run simulations, validate requirements, coordinate testing, manage regulatory workflows, and integrate across the full hardware development lifecycle. The cost of software creation collapsed once engineers could delegate implementation to AI systems. The same transition is now coming for the physical world. Flow is built for this shift. It exists to eliminate the operational bottlenecks between engineering intent and execution by creating an intelligence layer across complex hardware programs. It enables hardware engineering organizations to use agents to do the vast majority of their engineering work.

The result is a future where building a rocket, humanoid robot, or energy system starts to look less like traditional hardware engineering and more like programming software. This incredible value proposition has led Flow to partnerships with companies like Rivian, and Flow is now looking to scale their engineering team drastically in the next few months.

If you’re interested in changing how the physical world is built, come meet Pari at the next showcase.

Scott Nolan, Founder of General Matter

Right now, the primary bottleneck to scaling AI is energy, and unless you’re building data centers in space, nuclear is the answer. The problem is the U.S. lacks the fuel supply chain needed to support it. While electricity demand surges from rapid expansion of AI infrastructure and the push for reliable baseload power, the U.S. still lacks meaningful domestic capacity to enrich the uranium required for today’s reactors. America depends heavily on foreign suppliers for enriched uranium. That dependency has turned nuclear fuel into a strategic bottleneck for U.S. energy security, technological competitiveness, and long-term grid reliability.

Scott Nolan founded General Matter to rebuild this missing layer of American infrastructure. The team comes from SpaceX, Tesla, Anduril, and the DOD and approaches enrichment like a modern industrial technology problem: scalable, cost-competitive, and built for the energy demands of the next several decades. General Matter has already secured major Department of Energy support, including participation in the federal HALEU enrichment program and a $900 million DOE award to expand domestic enrichment capacity.

The next era of American energy security will not just require new reactors, but a fully rebuilt domestic fuel ecosystem that is scalable, reliable, and strategically independent. By modernizing enrichment infrastructure and bringing advanced fuel production back onshore, General Matter is likely to become foundational infrastructure for the future of U.S. energy.

Apply to attend the May 20th Startup Showcase and meet these founders in person.

Join the party and land that high-impact engineering or operator role before summer kicks off. Spots are already filling up so apply now to be considered for the showcase!

Next week, the following 10 founders will take the FYSK stage:

Rio Popper, founder of Conduit

Ian Livingstone, founder of Keycard

Neil Movva, founder of Sail Research

Jake Stauch, founder of Serval

Xerxes Libsch, founder of Specter

Paul Scherer, founder of Eigen

Scott Nolan, founder of General Matter

Andrew Lee, founder of Tasklet

Pari Singh, founder of Flow Engineering

Alex Atallah, founder of OpenRouter

A huge thank you to the Digital Ocean team for sponsoring the May showcase, and Rippling for hosting us again at their beautiful office!

Want to meet these breakout founders? Apply now to our showcase and to join our FYSK community.

We’ve reached a stage where the use of AI agents is widespread. They’re being deployed across industries, already working in sensitive systems, and running infrastructure overnight to keep companies operational. The question is no longer whether agents will run enterprise operations. It’s whether they’ll do it safely, efficiently, and in a cost-effective way.

While the AI labs continue pushing to make these agents more capable, engineers actually deploying them are no longer focused on the models alone. They’re focused on the deployment challenges, like giving an agent access to the right systems without risking a security breach and running thousands of agent calls overnight without spending a fortune on inference. These three companies at our next showcase are building the answers to those questions.

Neil Movva, Co-Founder of Sail Research

Neil Movva graduated from Stanford and spent time at NVIDIA and Apple, where he watched inference bills become the ceiling on what AI agents could actually do. Agents worth building — the ones doing deep research, running overnight code generation, processing data continuously — generate enormous compute bills. So companies don’t build them, or they build stripped-down versions that miss most of the value.

The open-source model ecosystem is competitive with frontier models on most real tasks, but the infrastructure to serve them reliably and cost effectively at production scale is lagging. That’s the gap Sail Research is looking to close.

Sail runs open-source models at up to 12x the cost efficiency of comparable proprietary alternatives and their benchmark performance — 82.4 on GPQA-Diamond — competes directly with existing models. The best part? Their API is OpenAI-compatible and they support leading open source models (e.g., Deepseek, Kimi, Qwen).

Sail is working from several angles, writing CUDA to increase performance on GPUs, maximizing efficiency of inference engines, distributing work, and pivoting between spot compute and reliable compute based on availability. In 2026, agents will be good enough to make meaningful, independent progress on hard problems, given sufficient tokens. Sail research will make sure they can.

Meet Neil at our next showcase to learn more about how his team are planning to dramatically increase intelligence per dollar, and make sure no compute is wasted.

Ian Livingstone, Co-Founder of Keycard

Through his work at Cape Privacy and Snyk, Ian watched enterprises give AI agents access to their most critical systems with no coherent security layer in between. At Cape Privacy, he tackled encrypted machine learning before most teams had figured out that data privacy and AI were on a collision course. He went on to scale Snyk’s developer platform from $30M to $300M in ARR, watching firsthand as engineering teams struggled to connect services and applications together securely. These experiences have led him to his newest problem: AI agents are beginning to operate across every system and organization, and nobody has built the security layer to handle them.

Agents are starting to work in production systems and operate autonomously across organizational boundaries, but the systems for allowing access were built for humans clicking buttons — static API keys, long-lived secrets, point-and-click authentication. None of that was designed for thousands of agents spinning up to complete a task and spinning down seconds later, each needing different permissions and access. Given the quick rise of AI agents, perhaps it isn’t surprising that even Slack and Salesforce ended up with leaks tied to agent vulnerabilities.

Keycard is looking to fix this. It’s a control plane for AI agents that cryptographically verifies an agent’s identity, enforces task-scoped permissions in real time, and maintains a full audit trail — without requiring developers to become security experts to use it. Instead of locking agents to narrow workflows or keeping humans in the loop for every sensitive action, Keycard issues short-lived, context-aware credentials that update dynamically as policies change. Access can be revoked instantly. The system integrates natively with Anthropic, OpenAI, and Microsoft, and Ian’s team is already contributing to the emerging standards (MCP, WIMSE, OAuth extensions for agents) that will govern how models securely connect with external tools and agents.

Ian co-founded Keycard with Matthew Creager, his former colleague at Snyk, and Jared Hanson, the creator of Passport.js and former Chief Architect at Auth0. Investors called it “the Auth0 moment for agent access.” Keycard is backed with $38M from a16z, boldstart, and Acrew Capital.

Learn more about Ian’s vision for the trust layer of the agent economy at our upcoming showcase.

Jake Stauch, Co-founder of Serval

Jake arrived at Serval after years of watching talented IT operators at some of the best-run companies in tech spend their days manually handling helpdesk work that should have been automated years ago. At Verkada where Jake led product, his customers were IT leaders who were motivated and desperate to automate. While these IT leaders loved what Jake was able to provide, it was clear they also wanted more. They were looking for a fix to an age old problem that until now required too many resources to effectively automate: the helpdesk.

Jake and his team developed Serval as the fix. IT teams describe their workflows in plain language, and Serval produces automations that are deployment-ready, traceable, and adjustable all the way down to code. Instead of a bandaid on top of legacy systems, Serval rebuilt the entire IT service management stack from scratch, with its own ticketing, access management, and asset management baked in.

What’s happened in practice has surprised even the founders. Customers haven’t just added Serval to their existing toolkit — they’ve ripped out their old platforms entirely. Automation has spread beyond IT into HR, Finance, Legal, and Engineering, with some teams abandoning dedicated software vendors because Serval handles the job better.

Sequoia, who led the Series B, said the customer conviction they’re hearing rivals what they saw when they first backed ServiceNow.

Join us at the showcase to see how Jake and his team are reworking IT from the ground up.

There’s a pattern across these three worth noting. Sail makes ambitious agents economically viable. Keycard provides the trust layer for enterprises to actually deploy them. And Serval is where that automation finally becomes something a non-engineer can own and operate.

The success of the agentic era won’t rely on the models alone. The tools to effectively deploy agents universally is the other half. Neil, Ian, and Jake are building these now. If you’re an engineer who wants to work on the hardest parts of the problem, this is a rare chance to hear directly from the people defining it.

Apply to attend our next SF Startup Showcase and meet these founders in person.

Something interesting showed up in the FYSK nominations this month. Some of the most compelling companies aren’t just based outside San Francisco — they’re building things that only make sense to operate outside it.

Here are a few that stood out:

• Halter (Auckland) — solar-powered GPS collars that let ranchers herd cattle from their phones. It sounds like software until you realize the product lives on actual cows, across ranches in Texas, Colorado, New Zealand. This only works if you’re close to that world.

• Starcloud (Redmond) — data centers in space. They’re putting NVIDIA H100s into orbit and used solar power and the vacuum of space for cooling. They moved to be near aerospace talent — the kind clustered around companies like SpaceX and Amazon — because that’s what the company actually depends on.

• Mariana Minerals (US, distributed) — part software company, part real-world operation. Coordination might sit in SF. But mining, logistics, and processing happen across Texas, Utah, and North Dakota because that’s where the resources are.

At first we thought these nominations might be early signs of a geographical shift away from SF as the center of gravity. But the more we sat with it, the more it felt like something else entirely.

For a long time, the most valuable startups were pure software. So they clustered in SF where the talent, capital, and network all lived. That hasn’t changed. What has changed is that, for many startups, software is no longer the whole product. It’s now the layer you apply to the real world.

Which opens up a different kind of opportunity:

Some of the most valuable companies now come from combining software with the physical world — in areas the frontier labs aren’t going to touch anytime soon.

A small team can now coordinate complex, real-world systems like farms, energy, supply chains, and satellites without needing massive scale on day one. You don’t have to own everything. You can integrate, orchestrate, and iterate.

At the same time, AI tools make it possible to build from anywhere. You can build with models like Claude or Codex anywhere in the world. The ability to build great software used to be tightly clustered. Now it’s widely accessible.

So you get this split:

• New capabilities are still created in places like SF

• But the application of those capabilities is spreading outward

Many of the opportunities to solve problems in the physical world aren’t ones the frontier AI labs will pursue anytime soon. OpenAI and Anthropic are focused on building foundational models. They’re not running mining operations, managing cattle, or launching data centers in space. Which means there’s this growing set of problems that benefit enormously from modern software and AI, require deep engagement with the physical world, and are still wide open for startups. Not because they’re more “defensible,” but because they’re underserved and increasingly tractable.

There’s a talent shift happening too. SF is still unmatched if you want to work on foundational AI alongside world-class researchers, with the highest comp, at the center of it all. But not everyone wants to build at that layer. Some engineers are increasingly drawn to building things that interact with the real world — agriculture, energy, infrastructure, aerospace. Places where the problems are messier, but the impact is massive and tangible. Those opportunities don’t all live in one city.

So this isn’t “SF vs everywhere else.” SF is still the hub where companies get started, capital concentrates, and ideas collide. It’s where the foundations of AI are getting built, one model at a time. But once you start building something that touches the real world, the map expands, because the real world has geography. It has cows, energy sources, mines, factories, and launch sites. If you want to build something that actually serves those customers, you don’t start in SoMa. You start where they are.

If you’re building something outside SF — or watching something big take shape somewhere unexpected — we’d love to hear about it. We’re always looking for standout founders to feature at the next Founders You Should Know showcase.

Looking to join a high-impact, fast-growing startup? RSVP for our next showcase (yes, in SF!) — we scour the world to bring you the good stuff.

We’re excited to announce three more upcoming dinners for engineers, all in San Francisco. Same format — a small table, no cameras, no decks, just a founder and a room full of interested engineers.

Early is everything and these are three companies worth knowing now.

The founders you’ll meet

These teams are growing quickly, and early hires will have real ownership and outsized impact.

Celeste Amadon, Co-Founder & CEO of Known

Known uses voice AI to replace the swipe — onboarding users through a real conversation, learning enough to make a single curated match, and getting two people to an actual date. In beta in San Francisco, 80% of introductions led to in-person meetings. Before Known, Celeste was a Stanford student working on AI applications.

Best for: engineers drawn to consumer AI with a genuine social mission and a product people actually love

Date: April 28th

Request to join here.

Adam Yala, Co-Founder & CEO of Voio

Voio builds a unified AI reading platform for radiologists — automating report generation across CT, MRI, and X-ray with accuracy that outperforms Google, Microsoft, and Alibaba’s leading models. Before founding Voio, Adam was an Assistant Professor of Computational Precision Health at UC Berkeley and UCSF, building and clinically validating AI models for cancer detection.

Best for: engineers excited about frontier AI with direct, measurable impact on human health

Date: April 30th

Request to join here.

Jake Stauch, Founder & CEO of Serval

Serval is an AI-native IT platform that automates help desk requests, access management, onboarding, and offboarding — replacing the legacy ticketing systems and drag-and-drop workflow builders that engineers have always hated. Before Serval, Jake was the director of product at Verkada.

Best for: engineers who want to build the new operating system for the enterprise

Date: May 7th

Request to join here.

The best founder conversations don’t happen on stage. They happen at a table, off the record, where you can dig into the details that never make it into a pitch deck.

This April, we’re bringing back our private dinner series, with intimate dinners where a small group of high-intent engineers and operators sit down with founders of inflecting companies to talk candidly about the real stories behind the build. These dinners are:

• Intimate by Design: We emphasize conversation over presentation. Small groups create space for probing questions, candid discussion, and a deeper understanding of the company’s inner workings.

• For Serious Explorers: These seats are reserved for engineers and operators who are serious about their next move and want to do their due diligence thoughtfully.

• Completely Off the Record: Everything discussed at these dinners stays at the table. These conversations won’t be recorded, shared, or posted anywhere.

Our first dinner in the April series will be with Airbase.

Airbase

The entire modern world runs on radio frequency spectrum from weather satellites and 5G to encrypted military comms. Every system we rely on competes for the same finite, invisible resource.

When the U.S. rolled out 5G networks, it nearly caused a catastrophic grounding of commercial aviation. Telecom companies had spent billions for new frequencies, but because federal agencies were operating with siloed data, nobody realized until the last minute that those new signals risked jamming the radar altimeters used by passenger jets to land. The result? Mass flight cancellations, international panic, and engineers scrambling to manually deconflict the airwaves at the eleventh hour.

Spectrum coordination still runs on static databases and PDF forms — tools built decades ago, operated by brilliant people who’ve simply never had anything better. This is the bottleneck Ari Rosner and Millen Anand set out to fix.

Airbase is the software platform for spectrum licensing, coordination, and intelligence, rebuilt for the pace of modern operators. The team is backed by a16z American Dynamism and Squadra Ventures.

The founders you’ll meet

Ari Rosner, CEO

Ari was employee #2 at True Anomaly, where he served as Chief Engineer on two orbital missions and scaled the company past 200 people. Before that he was a deep tech investor at Riot Ventures and mechanical engineer at NASA JPL — with hardware currently sitting on Mars.

Millen Anand, CTO

Millen led design and deployment of hundreds of next-generation Earth imaging satellites at Planet. He built RF payloads for geostationary satellites at Boeing, and was part of the inaugural class of a16z Engineering Fellows.

Interested?

The team is looking for software and RF engineers in New York City drawn to hard infrastructure problems with real stakes.

Request to join here.

We started Founders You Should Know with a simple idea: the best career-defining opportunities don’t come from job postings on Linkedin, they come from people. From being in the right room, at the right time, with the right builders. Or, as Josh Hoeflich, who dressed up as Aaron Burr from Hamilton for intermission show, would say, you want to be in “The Room Where it Happens.” However, if you missed the show last week, you can still catch a glimpse of what happened in our recap below.

The FYSK room is really special to me, because it’s built by my people. My brother came to the showcase. My husband emceed the show. The volunteers at registration are longtime members of this community — one of them is a founder we invested in who’s been coming since day one, and he brought his cousin along. And even our performer first came to FYSK as a jobseeker.

When you come to an FYSK showcase, you’re not just attending an event — you’re stepping into a community that compounds opportunities for you. We hope that some of our community members join the founders they meet at the show. Very often, what actually happens is you meet someone… and they introduce you to someone else… and then someone else. That n+1 effect is how both careers and companies get built.

I hope that through the FYSK community, you meet people who become collaborators, co-conspirators, and maybe even lifelong friends. If you’d like us to connect you with any of the founders below, send us an email blurb to forward!

Ricursive

Ricursive Intelligence is a frontier AI lab building a recursive self-improvement loop between AI systems and the chips that power them, compressing chip design timelines from years to weeks.

Founder: Anna Goldie

Open Roles: Engineering, Research, Security

Armadin

Armadin is a frontier AI security company building proactive security models and agents to protect organizations against attacks and maintain Western superiority in cyberspace

Founder: Travis Lanham

Open Roles: Research, Engineering, Design, Product, GTM

Zipline

Zipline is the world's largest and most experienced autonomous delivery service. With more than 130 million commercial autonomous miles flown to date, Zipline is transforming access to healthcare, consumer products, and food.

Founder: Keenan Wyrobek

Open roles: Engineering, Operations, Manufacturing + 150 roles!

Raindrop

Raindrop is an AI-native observability platform that helps teams monitor, evaluate, and debug AI agents in production.

Founder: Ben Hylak

Open Roles: ML Engineering, Product Engineering, Infra, Security, DevEx

TypeSafe AI

TypeSafe AI is building a new class of large language models designed explicitly for true no-human-in-the-loop automation, not conversational assistants.

Founder: Diogo Almeida

Open Roles: AI Research, Reliability Engineering, Generalist Engineering

Radical Numerics

Radical Numerics is using AI to master the language of life - making possible new medicines, and the design of life itself.

Founder: Eric Nguyen

Open Roles: AI and Bio

Fleet AI

Fleet AI builds simulated worlds and real-world challenge environments to help understand, evaluate, and shape advanced AI systems.

Founder: Nicolai Ouporov

Open Roles: Researchers, Environment engineers

Video not yet available

Deveillance

Deveillance is building privacy hardware that stops unwanted audio recordings, starting with Spectre I, a portable device that makes nearby microphones unable to capture intelligible speech.

Founder: Aida Baradari

Open roles: Hardware, Signal Processing, Engineering

Goodfire

Goodfire is a research lab dedicated to understanding and intentionally designing AI models.

Founder: Tom McGrath

Open Roles: Machine Learning, Research Science, Platform Engineering

Sunday Robotics

Sunday Robotics is building autonomous home robots that learn useful household tasks from large-scale real-world human demonstrations.

Founder: Tony Zhao

Open Roles: Full stack SWE, Product Managers, Robotics and ML

Want to be in the room where it happens for our next show in SF on May 20th? Apply now to our community and subscribe to our newsletter to be in the know about upcoming showcases.

Every time you make a phone call, stream a video, or board a plane, you’re depending on an invisible resource. And while over the last few decades we’ve moved from phones the size of a brick to devices in the palm of your hand that can stream orders of magnitude more data, the systems managing that underlying resource have remained unchanged.

Radio frequency spectrum is the medium beneath nearly every wireless signal. There is a fixed, finite amount of it available for different use cases – physics says so. You cannot make more. The system deciding who gets to use it, and when, and how much, was designed in the 1990s, and has barely changed since.

For a while, this was fine; the devices that needed spectrum were few enough that the patchwork held. Then satellites multiplied, drones went commercial, and autonomous vehicles started talking to infrastructure. The number of wireless devices on earth began doubling in ways that made the old growth curves look quaint, and spectrum infrastructure couldn’t keep up.

Millen Anand and Ari Rosner looked at this problem and came to a clear conclusion: this isn’t a policy problem – it’s a software one.

The Founders

Millen is the kind of person who will tell you, with complete sincerity, that he finds radio frequency allocation fascinating. At Planet Labs, he led a new class of Earth-imaging satellites, and the last thing he did before leaving was watch it go up on a SpaceX rocket.

That satellite was connected to Earth for approximately one hour a day. The rest of the time it was unreachable, flying dark, unable to send or receive anything. This is because connecting satellites to Earth requires radio spectrum and ground infrastructure, both of which are expensive and scarce. The industry has accepted this constraint as the cost of doing business in space, built their operations around it, and learned to live with the lag.

His cofounder Ari Rosner came in from a different angle. Ari was employee number two at True Anomaly, a defense tech startup where he wore every hat imaginable and grew the company from a few people in a co-working space to 200+ employees and over $400M raised. He left as Chief Engineer of two of their largest orbital programs.

Where Millen’s instinct was to go deep into the technical problem, Ari’s was to translate it: to walk into a room full of non-engineers and deeply understand their pain points and the urgency of the problems they were facing on a day-to-day basis.

Together, they shared two convictions: that the spectrum problem was solvable, and that nobody was seriously building the systems required to do so.

Pitching Washington

Driven by the conviction that the spectrum problem was a solvable software challenge, they came to Washington. What they did not have, starting out, was a company name or business cards. What opened doors was their framing: rather than arriving with complaints, as most industry contacts did, they showed up as engineers who wanted to be part of the solution. And rather than slide decks and vision statements, they showed up with things that worked, hearing about a problem on Friday and returning Monday with a working prototype.

Senior leaders in spectrum cleared their calendars. What Millen and Ari could articulate, in a way no one else had, was that they had a solution that could solve this massive problem, and deliver it in the timeline required to keep up with the exponential spectrum use. The regulators have the domain knowledge. They just lacked software capable of handling the scale and complexity of modern spectrum demand.

The timing is not incidental. A December executive order directed precisely this kind of modernization, the SAT Streamlining Act calls for automated satellite licensing, and the Golden Dome initiative has identified spectrum access as a critical military constraint. For perhaps the first time, policy is pulling in the same direction as technology.

Most companies emerging from stealth arrive with a story about what they intend to do. Last week, Airbase surfaced with something more rare. Instead of talking about what they will go build and what the end-user will be able to do, they’re already doing it. They already have a U.S. Government contract. And already have end-users actively testing their software.

The goal is not to wait. The goal is to have government regulators actively using the product in their daily work, so that the vision is not a promise but a fact.

Who They’re Looking For

Airbase is a team of seven people, and each of them owns something real. There is no orientation period, no ramp-up quarter, no safe lane to travel in while you figure out what you’re supposed to be doing. You will have an impact on day one.

The team is looking for software and RF engineers in New York City to build for customers like the NTIA, the FCC, the U.S. Military, and commercial spectrum users across the land, sea, air, and space domains.

More than any technical specialty, they’re looking for people who see something that shouldn’t work the way it does and can’t help but fix it.

Apply here: https://jobs.deel.com/airbase

The next wave of AI will bring extraordinary capabilities — and some truly unsettling ones. Systems that can design biology. Devices that can record and analyze conversations everywhere we go. These capabilities are arriving faster than most people expected, and they raise uncomfortable questions about privacy, safety, and control. But alongside those technologies, a new group of engineers is starting to build something just as important: the defenses.

Aida Baradari, founder of Deveillance

The year is 2028 and you’re sitting at a cafe doing an in-person meeting. The person you’re meeting with is wearing smart glasses that look completely normal. After the conversation ends you learn that everything you said was recorded, transcribed, and stored by an AI assistant running quietly in the background. You never agreed to it. You didn’t even know it was happening.

That world isn’t a science fiction future – it’s already happening. New wearable AI devices promise to capture and summarize entire conversations automatically. As speech recognition and large language models improve, your live conversations are turning into searchable, permanent data. What used to disappear at the end of a conversation can now live forever in someone’s cloud account.

That shift is exactly what Aida Bendari, the founder of Deveillance is pushing back against. The company is building a new category of counter-surveillance technology designed to give people control over whether their conversations become data. Their first device, Spectre I, detects nearby microphones and emits targeted cancellation signals that scramble what those microphones hear. To humans in the room, the conversation sounds completely normal. To recording devices, the speech becomes unintelligible.

The challenge is in having the system detect microphones, model how sound propagates through the environment, and generate interference signals precise enough to disrupt recording devices without disrupting the people actually speaking. It’s a problem that sits at the intersection of acoustics, signal processing, hardware design, and machine learning.

As AI systems get dramatically better at listening to and analyzing the world around us, technologies like this raise a deeper question. If every device can record everything, who gets to decide when it doesn’t?

Eric Nguyen, founder of Radical Numerics

A few years ago, a group of researchers at Stanford started asking a provocative question: what if the next breakthrough in biology didn’t come from a lab experiment, but from an AI model? That curiosity led Michael Poli and Eric Nguyen to build Evo, one of the first large generative models trained directly on genomic data. Instead of predicting or analyzing DNA, Evo could actually generate entirely new genetic sequences. It could design proteins and genomes that had never existed before, launching us into a future where AI systems help invent biological tools the same way generative models create images or code.

If AI can design biology, the upside is enormous. It could accelerate drug discovery, create microbes that break down pollution, or develop new treatments for disease. But it also raises a scarier question: what happens when the ability to design biological systems becomes widely accessible? The same technology can be used to design dangerous bioweapons.

Imagine a hospital lab in NYC detecting an unusual infection. The symptoms look like a flu, but the genetic sequence doesn’t match anything in existing databases. Normally it could take weeks for researchers to understand what they’re dealing with. But if the pathogen was engineered with AI, those weeks might be the difference between a contained incident and a global outbreak.

Now imagine having an AI system that can rapidly simulate how that organism behaves—how its proteins fold, how it interacts with human cells, how quickly it spreads. Within hours it can suggest possible countermeasures, predict mutations, and identify drugs that might neutralize it.

That kind of rapid simulation is one of the motivations behind the work at Radical Numerics. If our adversaries can use generative AI to design biology, we need equally powerful systems capable of understanding and defending against it.

Training models like Evo requires pushing the limits of compute, designing new kernels and architectures, and building systems that can simulate biological processes at enormous scale. It’s the kind of problem where advances in algorithms, infrastructure, and hardware all matter at once. If you’re curious about how AI might move beyond image and text generation to model and simulate biology, this is a rare opportunity to hear from the researchers helping define that frontier.

The technologies that Deveillance and Radical Numerics are building may seem like part of a Black Mirror episode. AI systems that can design biological molecules. Devices that can silently record every conversation around us. But the tools themselves are neither good nor bad, and the future isn’t predetermined. These powerful technologies can lead to surveillance and risk, or to safety and empowerment. The founders of Deveillance and Radical Numerics are working on very different problems, but they share the same instinct: if the future looks a little scary, the best response is to start building systems that make it safer.

Apply to meet Aida Baradari, the founder of Deveillace, and Eric Nguyen, the founder of Radical Numerics at our next showcase on March 25th in SF!

In the not-so-distant future AI systems will help decide which transactions clear in financial markets, how power flows across energy grids, and which medical cases get escalated to doctors. These systems are becoming powerful enough to control real infrastructure that affects every single person on the planet. The largest AI labs are racing to build more capable models, but they’re paying much less attention to the infrastructure needed to safely deploy them. Before AI can run the systems we depend on, we need ways to understand how it thinks, test how it behaves, and secure the environments it operates in.

Diogo Almeida, Founder of TypeSafe AI

You built an AI agent that’s supposed to process customer refunds. It reads the support ticket, looks up the order, and calls the refund API. The first few run smoothly. Then a ticket comes in that mentions a partial refund. The model decides it needs a helper function called calculate_partial_refund() and confidently calls it.

That function doesn’t exist.

Instead of stopping, the agent invents the logic, calls the wrong API endpoint, and refunds the entire order. Not once, but for every similar ticket that follows. Nothing crashed. The logs look fine. The system just quietly made the wrong decision. And it made it hundreds of times. These moments reveal the real challenge of this era of AI. The models are already powerful enough to automate meaningful work. But without reliability they’re useless.

That’s the problem TypeSafe is trying to solve. Instead of optimizing models for chat or creativity, the team is rebuilding the stack from first principles to create what they call transformative AI. Their model, System1, is designed for machine-to-machine execution rather than human conversation. The focus is reliability in environments where correctness and latency are non-negotiable: real-time decision systems, compliance pipelines, large-scale automation, or autonomous agents operating without a human constantly supervising.

Rather than bolting guardrails onto probabilistic models after the fact, TypeSafe is embedding reliability directly into the model itself. If that approach works, the payoff could be enormous. AI could finally take on the tedious but essential work that keeps companies running: compliance checks, operational workflows, data reconciliation, and thousands of small decisions that currently require human oversight.

The team behind the company makes this big vision feel achievable. Diogo Almeida, one of the founders of TypeSafe, previously worked at OpenAI, where he helped develop reinforcement learning with human feedback, the technique that made systems like ChatGPT actually follow human intent. That work became one of the most important breakthroughs in modern AI. Now Diogo is asking a different question. If the last generation of models learned how to talk to humans, what does the next generation need to look like if it’s going to run real systems? If you’re curious where frontier AI might go after the current wave of LLMs, this is a rare opportunity to hear from someone who helped shape the last one and is now building what comes next.

Nicolai Ouporov, Founder of Fleet AI

Imagine an AI agent helping coordinate activity inside a busy emergency room.

A patient is rushed in after a car accident. His blood pressure is falling. The triage nurse enters vitals, lab tests are ordered, and imaging needs to be scheduled immediately. Someone has to decide—fast—whether the patient should go straight to surgery, to a CT scanner, or to trauma observation. In a real hospital, those decisions happen in seconds and mistakes can cost a life.

In a simulation environment like the ones Fleet AI founder Nicolai Ouporov is building, an AI agent can practice these scenarios safely. The system might initially make the wrong call. It might send the patient to imaging instead of escalating the case to the trauma team. But the simulation shows what happens next—the patient deteriorates, alarms trigger, and the delay becomes visible in the outcome.

Because everything is simulated, the agent can run through thousands of emergency scenarios: internal bleeding, stroke symptoms, sepsis alerts, conflicting lab signals. Each time it sees the consequences of its decisions and adjusts its reasoning.

Over time, the system begins to recognize the subtle patterns that signal a life-threatening situation and learns when to escalate immediately. The goal isn’t to replace doctors. It’s to train AI systems in realistic environments where they can develop judgment before they are ever trusted with real patients.

That kind of training becomes essential as agents start helping with complex operational work. Before we trust software to coordinate systems that affect people’s health, finances, or safety, it needs the equivalent of residency training. Simulation environments give agents a place to build that experience.

For engineers, building those environments is a pretty cool problem. You’re creating digital worlds where intelligent systems can practice. That requires combining reinforcement learning, simulation, distributed systems, and developer tooling to design environments that are realistic enough to expose failure modes but flexible enough for agents to explore and improve. It also means thinking deeply about how to generate meaningful scenarios, measure progress, and surface the kinds of mistakes that only appear when systems interact in complicated ways. As agents start operating across thousands of tools and services on the internet, the need for this kind of infrastructure will only grow. Fleet sits right at that frontier.

Tom McGrath, Founder of Goodfire

In 2012, scientists realized that a strange molecular mechanism buried inside bacterial DNA could be turned into a powerful gene-editing tool. It was a total accident that they stumbled across this. No one was looking for it. That system became CRISPR, and it transformed biology almost overnight. What’s striking in hindsight is that the mechanism had been sitting in genomes for decades before anyone understood what it did. Today something similar may be happening inside AI models. Large neural networks are scanning enormous datasets in biology, chemistry, and medicine, learning patterns humans can’t easily see. But the internal reasoning behind those discoveries often remains opaque. If the next CRISPR-like insight is already hiding inside a model, we currently have no good way to find it.

While working as a researcher at Google DeepMind, Tom McGrath spent years studying how large neural networks actually work. What he kept running into was the same problem: even the people building the most advanced AI systems in the world didn’t really understand what was happening inside them. Modern models can reason, write code, and discover new patterns in science, but their internal decision-making still looks like a black box. McGrath became one of the early pioneers of a field called mechanistic interpretability, which tries to reverse-engineer neural networks and map the concepts inside them. After helping found the interpretability effort at DeepMind, he decided the problem was too important to stay purely academic and joined forces with Eric Ho and Dan Balsam to start Goodfire.

Goodfire is building tools to help researchers and engineers see inside modern AI models and better understand how they reason. Interpretability has been one of the most important and unsolved challenges in AI for years. If we want to trust these systems with meaningful tasks, we need ways to inspect their behavior, identify when something has gone wrong, and understand why a model made a particular decision. Goodfire’s work aims to turn what is currently a black box into something closer to a system engineers can actually debug. If we turn the Internet over to thousands of agents acting on our behalf, tools that help us understand what those systems are thinking will become an essential part of the stack.

The internet is turning into a place where software acts on our behalf. Agents will read documentation, move money, route resources, escalate medical cases, and coordinate complex systems we depend on every day. But before we trust them with that responsibility, we need an entirely new layer of infrastructure: models that behave reliably, environments where they can learn safely, and tools that let engineers understand what they’re doing.

That’s the layer TypeSafe, Fleet, and Goodfire are building. If you’re an engineer interested in developing the systems that will make AI trustworthy enough to run the world’s infrastructure, this is a rare chance to meet the founders defining this future.

Get excited to meet these three founders! Apply to attend our March 25th SF Startup Showcase.