Before Harsh Vardhan Goenka joined Kalpa Labs, he was at Rubrik Inc., where he helped build Rubrik Cloud Vault from its early days and rewrote the company's testing infrastructure so that test suites that took hours ran in minutes. He'd always gravitated toward small teams and hard problems — competitive programming took him to the ICPC World Finals, and engineering at scale brought him to Rubrik. But the pull toward something smaller, scrappier, and more frontier kept growing. When his former college seniors founded Kalpa Labs (YC F25), a company building generalist audio models, the timing finally clicked.

An 11 PM interview on a spare Windows laptop

Harsh knew the Kalpa Labs founders, Prashant and Gautam, from undergrad, where they'd co-founded a competitive programming club and spent hours debating algorithmic and mathematical problems. When Prashant first offered him a role, Harsh passed. But conversations continued, and four months later, he was looking for a change. Prashant asked if he'd like to interview for an open position.

"It was roughly 11 PM and all I had was a spare Windows laptop," Harsh says. "I ended up giving a 3 hour long interview, where I was given a problem around file system. I hadn't felt the same rush in the last few months, what I felt during that interview." He pauses. "All I feel it is God's plan."

What Kalpa Labs is building — and why it matters

Kalpa Labs is building generalist audio models. The vision is that audio hasn't had its GPT-3 moment yet. Rather than being limited to text-to-speech, the team wants one model that can voice a podcast from a script, sing a song in your voice, or generate audio that follows a detailed text prompt describing mood and tone. Some of those capabilities are already unlocked.

The coolest thing Harsh says he's built? "A voice model that sounds like human. Not kidding."

From infra engineer to ML researcher in five months

Harsh joined as Founding Engineer and Researcher, initially focused on ML infrastructure and software problems. But from day one, his founders pushed him toward ML research — a domain where he had no prior expertise.

"It has not always felt the best because there were times where I was clueless," he says. But that discomfort is part of the Kalpa Labs philosophy: everyone on the team is expected to stretch into unfamiliar territory. The thinking is that when someone builds expertise in a new domain, they bring a fresh perspective to hard problems that specialists might miss.

For Harsh, being an early engineer means refusing to limit yourself to engineering. "Have a great amount of ownership for the product — own different verticals, be it design, product workflows, anything that can help your product and user experience better. In this whole journey we somehow naturally push our limits that we have been limiting ourselves to."

Late nights, launch videos, and the cosine curve of startup life

One of his favorite moments had nothing to do with code. Three people in a room, brainstorming the script for their first launch video, recording takes, spending the night editing and hiding jokes in the final cut. "There were lot of cuts that didn't make it in the final video but was fun."

Working directly with the founders means discussing product direction, roadmap, hiring, and sales — and also game nights and house parties. Harsh appreciates that Prashant and Gautam are open about their own mistakes, talking through what went wrong and how to avoid it next time. "This removes an anxiety, fear, impostor syndrome within me that what if I end up making a huge mistake in anything I am working on."

He describes the emotional arc of startup life as a cosine curve: weeks where everything feels stuck, then a breakthrough that sends a dopamine rush through the team. "You need to recall all your wins and have the faith in each other that you will iterate and come out better."

How Harsh evaluates a startup — and who should join one

When someone asks how to evaluate a startup before saying yes, Harsh has a clear hierarchy:

• Who is on the team? — the most important factor.
• What are we building?
• How much runway do we have?

Harsh sees himself starting a company of his own someday, and the founders actively mentor him toward that. But for now, the problem at Kalpa Labs is too exciting to leave.

If you're an engineer who thrives on small teams, isn't afraid to be clueless for a while, and wants to help build generalist audio models at the frontier — Kalpa Labs is hiring.

Dylan Anderson was a software engineer at, in his words, "some random place in Minnesota." Then a friend-of-a-friend connection changed everything. A roommate of Adam's founders was asked if he knew any cracked engineers. Dylan's name came up. He did a work trial, moved to San Francisco, and became the first engineer at Adam (W25), a company building AI agents for CAD software.

Why he said yes

Dylan didn't need a long pitch. During the work trial, he noticed two things right away: the founders gave him real freedom and real trust from day one, and there was almost no friction in how the team worked together. "I recognized the talents of the founders that would make this work really well, each person complementing my own weaknesses and I complementing theirs," he says. He also saw clear, relentless drive from everyone involved. The move to SF was something he'd wanted anyway — to be at the center of things — and Adam gave him a reason to actually do it.

Building the engine, not fitting into one

As engineer number one, Dylan touches pretty much everything. The codebase has been rewritten multiple times. The systems in place are systems he built, shaped to his preferences and workflow. "If I want to build something to improve my life I do it, if something bothers me I fix it," he says.

He draws a sharp contrast with big-company engineering: "At a big company you're like a part of an engine, like an interchangeable piece that fits into a preexisting slot, you're an input to a function, whereas at Adam you're the one building the engine, you are the one molding everything to fit you."

His coolest build so far: adam.new.

Work as identity — without the burnout framing

Dylan is careful to say this isn't about bad work-life balance. It's that the boundary between work and life is blurred in a way that feels right. "I'm working on my passion, and I'm working with people that share the same passion," he says. The founders aren't just coworkers — they're some of his closest friends. That closeness raises the stakes on communication and teamwork, but it also means he cares more deeply about his craft than he ever could at a job he could mentally disconnect from. "It means my work is part of my identity in a more readily available way."

As Adam grows, Dylan expects his role to expand into mentorship and leadership. For now, the biggest shift has been internal: he's become more intentional, more value-driven. "Each of my actions directly impacts the performance of the company. Gone are the days of tinkering with something for purely academic purposes. Now I think about the value each thing I build provides."

His tools

• AI coding: Claude Code and Codex
• Editor: Zed (sometimes)

Advice for anyone considering an early startup role

Dylan's guidance is blunt: "Think about value deeply. Don't join a startup that's a tarpit idea. If you join and you put a lot of work into something that's a shitty idea, you'll hate everything you do and develop passive nihilism."

The right person for a role like this is someone who wants to own everything they build, cares about the problem more than the title, and is ready to let the work become a genuine part of who they are. If that sounds like you, Adam is hiring.

Andrew HunzekerHesed spent years building software seen by millions of people — food delivery logistics, CPG stocking at Walmart, video streaming at Major League Baseball, and most recently serving as Principal Engineer at Grindr. The work was fun and visible, but the general software development lifecycle was starting to look the same everywhere: "Exec to Design to PM to Engineer, with ever-increasing velocity due to AI." Then he met the founders of 10x Science (W26) at a talent summit, and something shifted.

A conversation that didn't sound like every other one in tech

10x Science creates AI-native software for scientific data analysis, currently focused on proteomics and drug discovery with plans to expand across the life sciences. But what drew Andrew in wasn't a pitch deck. It was how the founders talked.

"I talked to David about the beauty of math and working from first principles, and Vishnu talked about music, the science of sound, and how inefficient generative AI can be," he says. "It had been years since I had talked to people in the technology space who were not completely AI-pilled and were instead talking about how the principles behind AI can be used to create solutions instead of merely wrapping the commercial foundation models."

The interview process was rigorous — a multi-day work trial where Andrew had full access to the founders and they had access to him. When it ended, he still wanted to keep working on the problem they had given him "because it was so intriguing and their passion for the work so infectious."

All hands on deck, even hands that just showed up

Andrew's first week at 10x Science tested whether the startup energy was real. An opportunity came up to pitch a top-ten pharma company on a solution to a problem the team didn't even know existed. They had three days to bring Andrew up to speed on the problem space, design a solution, build it, and package it for a Friday demo.

"The thing we built has become a cornerstone in our tooling and has had applications outside of the initial pitch session," he says. "Other places I have worked would have pushed my start date back in order to just handle it on their own, but at a startup it is all hands on deck, even if those hands just showed up."

Finding The Thing, not just scaling the thing

For Andrew, being Employee #1 means something specific. It's not about scaling challenges — he's quick to point out that "there are thousands of ex-FAANG engineers who can help your company scale." What excites him is the earlier, harder phase: "Figuring out not the technical problems associated with the thing, but the actual thing itself. Do you have a hand in making the thing, or just in the scaling of the thing? The Thing, that is what being early means to me."

That discovery work has taken him places a big-company engineering role never would. At 10x Science, his name has appeared on posters presented at scientific symposiums, and he has submitted work to the American Chemical Society.

The company is now transitioning from exploratory research to locking down features for paying customers, which Andrew calls "really exciting." His tools are deliberately simple — the command line, VS Code, and whichever AI model "is the most SOTA that week." He bounces between providers and avoids getting sucked into prompt-tweaking rituals.

Who should join

Andrew's advice for anyone considering an early-stage role is direct:

• "Let your ego go, roll up your sleeves, and help out."
• "Will you get rich? Probably not. Will you gain more experience in six months than you can in years at a big company? Yes."
• "Life is too short to worry too much about these things, and life is too long to live with regret."

If you want to work with smart people on hard scientific problems, ship to real customers in the life sciences, and have the agency to build what needs to be built in the best way you know how — 10x Science is hiring.

Max Friedlander was building voice agents for service businesses when a LinkedIn message changed his trajectory. Ian, one of the founders of Petrarch, reached out to pick his brain about voice AI data. Max told them the idea wouldn't work. They kept talking anyway. When the team mentioned they had a YC interview coming up, Max asked them to keep him posted. A few weeks later, they got in — and a couple days after that, they asked if he wanted to be an intern. "Just based on the vibes from the call we had before," Max says.

He was employee number five. He's 20 years old, an applied mathematics major at Middlebury College, and he does sales at Petrarch, a YC S26 company that supplies data for frontier AI labs.

Why he said yes

Max had tried starting his own ventures before and been stood up at the last minute for other opportunities. He had traditional internship offers on the table — SDR and BD roles at bigger companies — but they felt wrong. "What felt safe began to feel suffocating and limiting," he says. "I wanted to be pushed and flexible, seeing what I could learn."

The deciding factor was the founders themselves. They were driven and ambitious, but also people he'd want to hang out with outside of work. "When it comes to evaluating whether the company is the right choice, it's more of a gut feeling than anything else, especially when it comes to people."

Leading deals with complete autonomy

Within his first months, Max was leading his own deals and negotiating high-stakes contracts — the kind of responsibility that wouldn't come for years at a large company. He handles legal review, sales, and communication with all parties on his own.

The reality is less glamorous than it sounds. One night, the team believed they had a multiple five-figure deal secured. Everything fell apart at the last minute. Max had to restructure the entire deal within 36 hours and secure three signatures. "It was challenging, but it was also incredibly rewarding to have such autonomy and the freedom to make my own decisions," he says.

He's also built tools to support his work — a scraper and harness to find obscure leads and automate outreach so prospects book directly into his calendar. His stack includes Claude Code in the terminal, Apollo, and VS Code.

The highs and lows are real

Max is candid about what startup life actually feels like. "It's really hard to be level," he says. "The highs and lows are emotionally jarring, a lot more than any other type of job I have worked in, for better or worse."

What surprised him most was how much his strategic ideas mattered. The founders listen. There are no strange dynamics around seniority — everyone on the team is the same age, and Max describes the culture as one where people can be serious when they need to be but still crack jokes and be themselves. "In a startup you spend a ton of time with the team, even more than with a partner or significant other," he says. "That's key to success."

The experience has changed him in a specific way: confidence. "The more you do, the less afraid you are to act like a founder."

Who would thrive here

Max plans to start his own company one day. For now, he has a clear message for anyone weighing whether to join a startup early:

"It's hard, rewarding, fast, and you will act like a founder more than an employee. You need to be a fast learner and super curious. It will ultimately condition you to be a thinker, not just a worker."

If you want autonomy over real deals, a team that treats you like a co-founder, and a front-row seat to the AI data business, Petrarch is hiring.

David Mount has started four companies, served as CTO, watched funding dry up, and built a passion project about nuclear energy called Atomic Ambitions. When his previous startup, Chinchill.AI, ran out of money, he signed up on YC's Work at a Startup. Two weeks later, a founder from 10x Science reached out — someone who had actually taken the time to experience Atomic Ambitions. That detail mattered.

10x Science (W26) uses AI to process scientific data and unlock bio-chemical discovery, with work already proven in cellular-scale advances across medicine, biotech, and pharmacology.

A work trial that felt like a final exam

The interviews clicked, and Mount was invited for a two-day, on-site work trial. The problem he was given to solve was, in his words, "more than a stretch." After a day of research and design, he spent the night cramming proteomics, deep AI theory, and new frameworks in a proprietary application. On the second day, he explained to Codex what to build and cranked away until time was up.

"I felt like I was back at college going for as much partial credit as I could get, while having to face the fact that I hadn't quite measured up," he says. "Or had I? Perhaps the best measure comes from a challenge that pushes someone to the limit."

Full stack, zero delegation

Mount is employee number two — a position he has held before at another company, which earns him the self-appointed title of "Elder Statesman." At a big company, he says, he would be a VP or CTO: lots of responsibility, all delegation, no hands-on action. At 10x Science, he works directly on anything and everything:

• Full-stack development, mostly Rust and Vue.js
• Zero-to-one features and rapid improvements
• Agent orchestration
• Benchmark runners and a showcase for newcomers who want to run scientific analysis routines in isolation and learn the associated science and math
• The build process itself

He is still on a steep learning curve and says one of his coolest accomplishments so far is learning to identify the twenty amino acids found in humans. "Neat," he adds, without irony.

Why this company, why now

Mount evaluated 10x Science the way he evaluates any opportunity — with a short checklist:

• Is this company helping humanity?
• Are the people smart, upbeat, driven, friendly?
• Is the product idea real enough? Is the timing favorable? Is the field green and open?
• Do they like me?

"The founders made it easy. Everything they said checked the boxes for me. Their rigorous yet compassionate evaluation of me sealed the deal."

He describes the founders' backgrounds as uncommonly varied — one from the Bahamas, one a product of Silicon Valley, one from North Carolina, all converging in a research lab at Stanford. Their first hire was a circus performer before jumping into software. "10x Science is assembling quite the crew for this grand adventure. Real diversity."

Working with the founders

Mount calls the experience of working directly with the founders "amazing." They are open, visionary, and productive. "Assignments feel more like opportunities. We get all of the rope we need along with a safety harness, the sense that nothing can go too far off of the rails. We're building a warp drive, so to speak."

His setup reflects the company's willingness to invest in its engineers: a top-of-the-line MacBook Pro, a Keychron Q1 Max keyboard ("for that extra clicky-clacky typing"), an ultrawide monitor, and unrestricted access to Claude, Codex, and whatever models help maximize leverage.

Who would thrive here

Mount has been through a dozen startups. He has no illusions. His advice for anyone considering an early role: "Learn to be okay with ambiguity. Forget about work-life balance. Work in bursts. Take breaks as needed. Stay fluid. Try ideas quickly. Shed whatever is not working. Polish nothing. Despair leads to inspiration. Enjoy the ride."

If you want to help scientists searching for cures, are comfortable building across the entire stack, and can stomach a two-day work trial that might keep you up studying proteomics — 10x Science is hiring.

Yair Alon spent a decade at Google Research and DeepMind. He built the deep learning models behind Portrait Mode on Pixel phones. He published papers on efficient deep learning. He designed models for Search, Ads, and YouTube. His last project was VideoPoet, a video generation system. Everything was great, by his own account. But he felt that small startups had become where the most exciting innovations were happening. So when former DeepMind collaborators told him they were starting a company, he knew immediately he wanted in.

That company is Poetiq (YC S26, San Francisco). In one sentence: Poetiq is building an AI system that can optimize any other AI system, including itself.

Leaving Google after a decade

Yair had been at Google since his PhD internship at CMU, where he studied machine learning and computer vision. He knew how to be productive and successful inside that ecosystem. Leaving felt scary. "I was there for so many years," he says. "I knew how to be productive and successful in that environment so leaving felt like a scary new thing." He already knew the founders and trusted the technical direction, so the real hurdle was personal. He had to find the courage to walk away from something comfortable toward something unproven.

What he actually does all day

Yair is one of a handful of people who started on the first day of the company. His official title is Founding Research Scientist. Unofficially, he calls himself Coffee Evangelist. In practice, being early means working on everything, sometimes in a single day: "You could be working on a new ML technique in the morning, fixing issues with the DB later in the day and then in the afternoon meeting a realtor to look at a potential new office."

One project he is particularly proud of is the experimentation infrastructure. At Google, entire teams handled experiment launchers, dashboards, and visualization. At Poetiq, none of that existed, and the work is novel enough that off-the-shelf tools were not relevant. He and the team built a first-tier experimentation system from scratch, with detailed analysis of how their optimizer is performing.

Building an AI that improves itself

Poetiq's core bet is on recursive self-improvement: an optimizer that can improve any AI system — harnesses, agents, skills — and that can also improve itself, so the rate of improvement accelerates. Yair puts it simply: "The main thing we are building at Poetiq is an AI that can improve other A.I.s and itself. I think nothing can top that."

Early on, the work was split between two tracks: figuring out the algorithmic foundations of the meta-optimizer and building the experiment-running infrastructure to validate it quickly. More recently, his focus has shifted to scaling — identifying what is missing so the system can run longer without human intervention, finding what would cause friction if they wanted to ten-times their experiments, and spending significant time on hiring and prototyping Poetiq's interview process.

People, challenge, autonomy

Yair describes three things that matter most to him in a company: people, challenge, and autonomy. On people, he says the team is genuinely enjoyable to spend time with. On challenge, he frames it as the rare combination of creating a new domain of AI research while simultaneously inventing a new product category. On autonomy, the founding team — all with deep experience in industry and academic research — avoids micromanagement. Instead, they run frequent deep-dive syncs to stay aligned on priorities and converge quickly when needed.

He is candid about tools, too. He has used VS Code, Anti Gravity, and now mostly Claude Code, but he thinks people put too much emphasis on the setup. "The key thing is knowing what you want to build, having a clear design for it in your mind, and knowing what success looks like. Then whether you use Claude Code, Codex or whatever doesn't matter that much."

Who would thrive here

Yair's advice for anyone considering an early-stage role is grounded: "Realistically most startups do not become Anthropic. So you need to go into it with the correct mind set. The goal is to learn a lot, be part of a small team executing together, and be able to contribute across the board. Anything more is just icing on the cake." If you are an engineer or researcher who wants ownership over the full lifecycle of what you build — and you are drawn to a problem no one has solved before — Poetiq is hiring.

When Varsha Parthasarathy was pregnant, she needed a letter from her doctor's office just to use her own HSA money on pregnancy-related expenses. Everything was already in her chart. She still had to send her husband or her sibling to the admin office to chase down the paperwork. Four months after her first kid was born, running on very little sleep and definitely not planning a career move, she sat down with her former Lyft colleague Miguel — one of the cofounders of Flex (S23) — to work together for a week in San Francisco. By the end of that week, she was in.

Flex helps businesses accept HSA and FSA payments, so people can use their pre-tax health dollars on products and services they're already buying. It plugs into a merchant's existing checkout and shows up as another way to pay, like Klarna or Google Pay.

From self-driving cars to healthcare payments

Before Flex, Varsha worked at Lyft's self-driving division on the autonomy side — data annotation pipelines, supervised metrics, sensor data stores, offline ground-truth perception. She also spent a couple of years as an engineering manager there, which taught her that the people matter the most.

She already knew Miguel as a coworker. Then Sam, Flex's other cofounder, reached out as a reference call that turned into something much longer. "I came away really liking Sam," she says. But she also knew that working with someone at a big company is different from building something from scratch with a handful of people. So she tested it: a week side by side, watching how Miguel thought through problems and made decisions. "By the end of that week, I just knew these were people I wanted to build with."

What being engineer #1 actually looks like

Varsha was Flex's first engineer. She doesn't have a formal title. If pressed, she describes herself as "the engineer who also does support, analyzes queue failures and metrics, and every now and then jumps on a sales call or helps onboard a merchant."

She still remembers the company's first Thanksgiving. She was on call for both engineering and support — answering questions from customers and merchants having trouble checking out while simultaneously chasing down a decimal computation bug. "In that one weekend, I touched more parts of the product than I would have in months at a big company, and I loved every bit of it."

Sales calls, merchant onboarding, content pipelines, marketing compliance — none of it was in her job description. None of it existed before she did it.

A Slack bot that became Flex's first line of triage

On-call comes around often on a small engineering team, and Varsha wanted to spend less time on support and more time building. So she started writing small bots. One was a Slack bot that helped her figure out what was going on when a payment failed. Over time, it grew into the Flex Investigator. Give it an order number or a customer email, and it traces what happened across Flex's systems in about fifteen seconds — something that used to take around twenty minutes. Today it's the team's first line of on-call triage.

Nobody had to sign off on it first. "Sam and Miguel trust us to go build what the team needs, and we deliver."

She also built a personal agentic bot that lives behind a Slack emoji: an agent that listens for events, finds answers to questions people ask her, and keeps working around the clock, even pushing PRs. On the tools side, she reaches for Claude Code for implementation, Codex for brainstorming and planning, and a personal finance bot she built and named Paisa Bot — paisa means money in her language — to track markets and think through investing strategies.

Founders who are "very human"

Varsha says anyone on the team can go to Sam and Miguel with uncomfortable or awkward questions, directly, and they actually want that. "It goes both ways. They've given the team a lot of trust, and we trust them right back. Whenever those questions come up, there's always some action that comes out of it. It never just ends with a conversation."

What surprised her most about startup life: she never felt like she had to choose between work and her kid. "And the founders — I didn't expect them to be this fun and chill. They're very human, and that makes them so easy to work with."

Who would thrive here

Varsha's advice is specific:

• If you enjoy solving problems across different domains and don't have much ego about which problems you work on, a startup is a great space for you.
• Pick a company where you truly believe in the mission. "There aren't many companies working where payments and healthcare meet, and building something that keeps getting better for both customers and merchants is far from easy. There are still so many hard problems left to solve, and that's exactly the kind of environment I enjoy."

Before he ever wrote a line of code for a startup, Sherif Amer was designing embedded systems and verifying mixed-signal chips at Infineon. Then an unexpected layoff changed his trajectory. Rather than jumping back into semiconductors, he spent two and a half years traveling the world, teaching himself software engineering, and building his own product — a project called Roamies. That winding path eventually brought him to Indemni (W24), a San Francisco startup that combines driver verification, carrier intelligence, and real-time shipment monitoring to help freight brokers detect fraud and move loads with confidence. Sherif was employee #1.

From hardware to founding engineer

Sherif's official title is Founding Engineer, but he describes himself as "a product-minded engineer who happens to write a lot of code." His background in electronics gave him a deep instinct for understanding how things work from the ground up — an instinct he now applies to software architecture, product design, and everything in between.

When he evaluated Indemni before saying yes, he wasn't chasing a title. "I cared less about titles and more about the people I'd work with, the problems we'd solve, and how much ownership I'd have," he says. He wanted founders who welcomed technical disagreements and trusted engineers to contribute beyond writing code. He found that.

Building the infrastructure from scratch

Sherif designed and built much of Indemni's asynchronous processing and notification infrastructure — Redis Streams consumers, message recovery, concurrency controls, rate limiting, and delivery. He's also built out the verification infrastructure, connecting external identity providers, orchestrating multi-step workflows, and designing how the company processes and analyzes verification results.

"Rather than automatically reaching for a managed cloud service for every problem, I got to evaluate the trade-offs and build solutions that made sense for our scale and needs," he says. The range is wide: one day it's distributed systems and reliability, the next it's database models, analytics dashboards, or a customer-facing feature. "And when something breaks, there's no mysterious infrastructure team to hand it off to. That's usually my problem too."

One project he's especially proud of: building a job scheduler by extending Indemni's existing Redis Streams infrastructure. He used an AI coding agent as a collaborator, providing architectural context and constraints while delegating implementation. What made it satisfying wasn't that AI wrote the code — it was that he could approach the problem as an architect, choosing the technologies, defining the trade-offs, and reviewing the result.

Autonomy, accountability, and a lot of rope

"A lot of rope!" is how Sherif describes the working relationship with Indemni's founders. He's not expected to just agree with an idea and go implement it. "We challenge each other, debate approaches, and sometimes disagree quite strongly, but there's always room for an honest conversation." Those conversations aren't limited to engineering — he can question product decisions, suggest new directions, or flag opportunities the team hasn't considered.

As Indemni has grown and more engineers have joined, Sherif's role has shifted. He now spends more time on architecture, establishing engineering guidelines and code quality standards, and mentoring newer team members. "One of the most rewarding changes has been seeing those engineers become increasingly independent," he says. "They now make solid technical decisions and ship quality work without needing my input at every step."

What surprised him most

Coming from a large semiconductor company with layers of review and clearly defined responsibilities, Sherif was struck by how much startup life revolves around making decisions with incomplete information. "You have to be comfortable making reasonable decisions, moving quickly, and adjusting as you learn more," he says. "I've come to appreciate that a lot."

He was also surprised by how much trust the founders placed in him — not just over engineering, but over product direction and company decisions. "I expected to have ownership as a founding engineer, but I didn't necessarily expect to have such a strong voice in decisions beyond engineering."

Who would thrive here

Sherif's advice for anyone considering an early-stage role is direct: be honest with yourself about what you want. The work is rewarding, but only if you're comfortable with uncertainty and figuring things out without a guide.

• You won't always have perfect requirements or established processes.
• You might design a system in the morning, implement it in the afternoon, and troubleshoot it in production the next day.
• You'll get exposure to parts of engineering and business that might take years to experience at a larger company.

If that sounds like the job you've been looking for, Indemni is hiring.

Before Alex Lee joined Magic Patterns (W23), he was running his own company, Iago — one that's still self-running today. But when his former Robinhood colleague Teddy co-founded Magic Patterns, a tool that helps product designers create interactive mockups and prototypes, Alex saw something he couldn't pass up: an AI-first approach to a problem he'd hit himself while building new features at Iago. "It's pretty clear that Figma is going to be replaced and a company like Magic Patterns is here to replace it," he says.

Technically, Alex was employee number two — a sales hire came first — but he was already working part-time before he officially started.

Why Magic Patterns

Alex and Teddy stayed in touch after Robinhood, texting about startups and sharing philosophies about work and engineering. "I felt like I aligned really well with Teddy and Alex (Danilowicz)" — the co-founders — "and how they thought, and felt like I'd not only grow but grow with the company if I joined."

The problem space pulled him in, too. Magic Patterns is used by more than 3,000 product teams, and Alex gets a front-row seat to how those teams are reshaping design in an AI world. "Especially as the different product team roles — eng, design, product — start to blend, it's super interesting to see how fast everything's evolved, and stay on top of it."

Owning problems, not just features

"You're not just a code monkey, you're a product person," Alex says. "Things move so fast at a small company that you don't just own features, you own problems. And when you find solutions, you're in charge of finding the next set of problems to solve."

That means coding, yes — but also improving the onboarding funnel, talking to customers, and jumping on support. One weekend, Alex noticed users weren't just making designs but building slide decks to present those designs to leadership. So he built out an entirely new file type with full slide deck features over the weekend and shipped it. "A big company wouldn't even let me ship this without talking to the right team who 'owned' this product surface."

He also landed on the front page of Hacker News with a side project done at work: an agent that builds Rollercoaster Tycoon-influenced theme parks from a prompt.

The daily toolkit

Alex dogfoods Magic Patterns itself to design new features. Once something is ready to implement, his workflow looks like this:

• Cursor and Claude Code, always on the latest models
• Cursor connected to Slack — he prompts it directly for small bug fixes and features
• An internal code reviewer agent called Patterson that reviews code before it ships
• Three clones of the same GitHub repo instead of worktrees, so he can work on features in parallel

Transparency as a feature

Alex says the thing that surprised him most about startup life is how transparent it is. "Everyone at Magic Patterns knows the financials, growth rate because the founders trust us, but also because they are building a very open and transparent culture." He actually thinks large companies carry more hidden risk. "You never know when they might lay people off or have a change in direction."

The founders, Teddy and Alex D., foster a culture of trust and autonomy. "They're incredibly receptive to feedback and will always take action on it."

Now that the team has grown to ten people, Alex has taken on more interviewing and has become the dedicated forward-deployed engineer for enterprise customers, especially around Magic Patterns' Design Systems vertical — the top-requested enterprise feature.

Who should join

"If you're early in your career, make a polarized bet," Alex says. "You'll work more than a typical 9-5 but you'll be learning a lot more in return." If you want real ownership, the chance to ship a feature over a weekend, and a view overlooking Coit Tower from a new San Francisco office, Magic Patterns is hiring an engineer who wants to build the world's best AI design agents.

Vin Bui was on a leave of absence from Cornell when he found Nowadays. He had just left his own YC company from the Winter 2025 batch and was back on campus, not enrolled, scrolling through Work at a Startup looking for what came next. A dog logo caught his eye. "It felt childlike in a way, and that unique taste is what originally attracted me," he says. Then he looked at the website. "This website needed a lot more work. I really like seeing companies that have things I know I could fix."

He sent a message to co-founder Anna, went through the interview process, and became the first full-time hire at Nowadays (S23), a San Francisco company that uses AI to make planning corporate events simple. When it started, Nowadays was selling to small startups. Today it’s the platform Fortune 500 companies like Google, Airbnb, and BlackRock use to bring their people together.

Acting as the technical founder

Vin joined the company as employee number one, and the first technical hire. "As engineering hire #1, you’re basically a technical founder," he says. "Everything I do is exactly what I would've done if I were the technical founder of my own company."

That meant building engineering from the bottom up, setting up processes, leading SOC 2 compliance, handling security, and becoming the main point of contact for anything touching engineering or IT. He talked with heads of engineering at other companies to figure out what good looked like, then built it himself. "No mentorship exists, right? You have to mentor yourself."

Why he said yes

Vin evaluated Nowadays the way a founder would evaluate a co-founder. He looked at two things:

• Could he have impact? During the work trial, he immediately saw what the codebase needed. He had the background to fix it and knew he could build something meaningful.
• Could the founders make the company succeed? Having been a YC founder himself, he watched how Amy and Anna worked — how they presented the company, attracted customers, and communicated their plans. "These two know exactly how to present the company. Working with them, I understood how competent they are and that I could rely on them."

He also asked them directly about their plan for the future. "The way they answer and the way they sound tells you whether they've actually thought it through. I really like people who think."

Building AI voice — and everything else

When asked about the coolest thing he has built at Nowadays, Vin lights up about AI voice: building systems that call hotels and negotiate with them on behalf of clients. "Voice is very similar to how humans work, so it's very interesting to see how close AI can get to us."

His daily setup reflects someone who cares about craft. He uses Ghostty for his terminal, Claude Code CLI over any GUI, and orchestrates different AI models depending on the task. "I'm not the type to use the biggest model for everything. Using the right model for the right thing matters a lot." He writes his own custom skills and workflows, tests every new model that comes out, and browses Twitter to find tools worth trying. "I love building my own workflow."

At a big company, he says, you might spot something great for the product but can't act on it. At Nowadays, it's different. "I observed something and have a great idea, let me just build it."

From builder to the person the team bounces ideas off of

Nowadays is now an 11-person team, and Vin's role has evolved. He still builds, but he has become the first person teammates come to when they have a big idea they haven't fully thought through. "I wouldn't call myself a manager, because I don't manage anyone. I trust them and give them full autonomy." He triages tricky problems, then hands them off to the right person to go deeper.

He is also quick to point out something that surprised him about startup life: you can have a life outside of work. "Amy has a kid, and Anna loves exercising and has been running marathons. We're a health-first type of team, and we care about each other." He values that the founders work hard and still make time for families and friends.

Who would thrive here

Vin is direct about what early startup life demands. "You need to be able to trust yourself, because the main person who can help you is yourself. You need to be self-reliant and constantly learning on your own." But it should also feel good. "You spend so much time with them that it should feel like being back in college, working on something with your friends."

If you want the intensity of a founding role without investor talks, if you want to pick your own problems and ship fast, and if you want to work with people you would genuinely spend time with outside of work — Nowadays is hiring.

Ryder Kemper was in Ohio, winding down a startup he and his college friends had built together. The team had decided to close it — not because it failed, but because they wanted different things. Ryder knew he wanted to stay in startups, and he knew he wanted to be in San Francisco. A message on Work at a Startup from Corgi Insurance's first sales hire changed everything. He did a thirty-minute interview on a Friday, flew to SF on Sunday for a work trial, and never went back. "By the time I had an offer, I had already helped to get the signature Costco mattresses into the house we were living in."

Corgi Insurance (YC S24) is rebuilding core financial infrastructure from the ground up — insurance systems, policy administration, claims — with the goal of owning the stack end to end.

The Mystery Machine test

Ryder's work trial was not a whiteboard exercise. He spent roughly half his time writing code and the other half riding in the company's Mystery Machine bus, hauling furniture to a new office. That ratio told him something important about the kind of place Corgi was. "The energy was just different. I felt like there was a good chance of success but worst case scenario it would be a great time."

He calls himself employee number seven. His self-assigned title: Chief Furniture Mover.

What he actually builds

Ryder has touched nearly every part of Corgi's technical stack:

• Core policy administration and claims systems
• New product development as Corgi launches additional insurance lines
• Hiring processes for new engineers and teams

The claims system is the project he talks about most. Working closely with adjusters who "think its all magic and crazy how fast we can put stuff together," he gets fast, enthusiastic feedback that makes the next iteration easier to ship. The loop is tight: build, get it live — sometimes before it feels fully ready — then iterate on real feedback from customers and the sales team.

His current tool of choice is Claude Code. He recently set up a desktop at home he can shell into remotely, letting his AI agents run even when his laptop is offline. One trick he likes: having Claude generate artifacts beyond terminal output — diagrams in Figma, recorded videos of pull-request work — to make its output more understandable to the rest of the team.

Why the people matter

"The concentration of talent here is really crazy," Ryder says. What stands out to him is the default posture: no idea is too big, no challenge gets waved off. "The default answer is always 'yea we can do that,' or 'why don't we just do it ourselves.'" The founders give people real freedom to operate, and Ryder says he has learned a lot about business just by watching how they navigate situations nobody on the team has seen before.

He also notes, with equal sincerity, that he was surprised by how much milk co-founder Nico can drink.

Two years, ten years of work

Ryder estimates he has packed about ten years' worth of experience into roughly two. He has shifted between building, managing, and scoping new projects, but has stayed close to whatever Corgi is working on next. "To me everything still feels like a new startup." He has also discovered something unexpected about himself: he is happier with less stuff. Moving around frequently for work, he stopped accumulating things and found he didn't miss them.

He plans to start his own company one day. For now, Corgi is the place where he is learning how.

Who would thrive here

If you want clear boundaries and a narrow job description, this is not the role. If you want to ship fast, own entire projects end to end, and do whatever needs doing — engineering or otherwise — Ryder's advice is simple: "Expect a lot of uncertainty but also a crazy journey that has not really been matched by anything else in my experience. Expect to work extremely hard and constantly feel like there's never enough time for anything. Expect to learn a lot about yourself and make a new group of close friends."

Corgi Insurance is hiring. You can find them on Work at a Startup.

We’re thrilled to announce that Vivian Midha Shen and Raphael Schaad are joining Y Combinator as our newest General Partners.

Vivian and Raphael have spent the last two batches working closely with YC founders as Visiting Partners. We’ve seen firsthand what makes each of them exceptional: they combine deep technical experience with strong product judgment, and they know how to help ambitious founders turn an early idea into something users love.

Vivian co-founded Juni Learning (YC W18), an online academy for kids’ coding and math, and later founded Acely, an AI tutor for college readiness. Across the two companies, she built businesses with more than $50 million in revenue and millions of users. Acely alone grew from zero to $10 million in run rate with a lean team, showing what small, AI-native companies can accomplish.

Vivian has done nearly every job inside a startup: engineering, product, design, marketing, and whatever else needed to get done. Before becoming a founder, she worked as a software engineer at Google, a consultant at McKinsey, and Director of Product at Operator. She studied computer science and creative writing at Stanford. That range helps her judge both how well a product works and how much it matters to users.

Raphael founded Cron (YC W20), a next-generation calendar beloved for its craft, speed, and attention to detail. Notion acquired Cron in 2022, and Raphael went on to grow Notion Calendar to millions of users. He then served as Head of Design for Growth and new products.

Raphael moves fluently between design and engineering. Before Cron, he worked on the popular news reader Flipboard and the award-winning Markdown editor iA Writer. With his Swiss eye for design, he has spent his career making complex software simple. Raphael holds a BS in Computer Science from Bern University of Applied Sciences and an MS in Media Arts and Sciences from MIT. Founders building new products benefit from his instinct for what matters and his insistence on craft.

Vivian and Raphael have built companies from the ground up, and they’ve already shown how much they can help other founders do the same. As General Partners, they’ll work with companies from their first weeks in the batch through the years of building that follow.

We couldn’t be more excited to have them on the team. Welcome, Vivian and Raphael!

We're excited to announce that Christopher Golda and Grey Baker are joining Y Combinator as General Partners.

Chris and Grey have both been Visiting Partners at YC, where they've already made a big impact working closely with founders across multiple batches. Now, as General Partners, they'll play an even bigger role in selecting, advising, and supporting the companies we back.

Chris co-founded BackType (YC S08), a data infrastructure and analytics company acquired by Twitter in 2011. At Twitter, he launched and ran the Ad Center product, growing revenue to over $1B. Since leaving Twitter, he's been working directly with founders to help them find customers and raise capital, participating in early-stage rounds of companies like Benchling, Coinbase, Stoke, and Supabase. As a Visiting Partner, he's brought that combination of deep technical founder experience and a decade of pattern-matching across hundreds of early-stage companies to the founders he works with.

Grey co-founded Dependabot, a developer tool used by over a million people that was acquired by GitHub in 2019. Before that, he was an early employee at GoCardless (YC S11), joining when it was just six people and leading product and engineering as the company grew to over 100. After GoCardless, Grey cycled 29,000 km around the world, then came back and built Dependabot as a side project that took on a life of its own. He went on to co-found Pincites (YC S23), which was acquired by Filevine. As a Visiting Partner, he's brought a builder's mentality and firsthand experience scaling products from side project to millions of users to every batch he's been part of.

As General Partners, Chris and Grey will work directly with YC founders at every stage of their companies. We're super excited to have them on the team. Welcome, Chris and Grey!

We're thrilled to announce that Diana Hu has been promoted to Managing Partner at Y Combinator.

Diana first came through YC as a founder. She co-founded Escher Reality (S17), an augmented reality backend that was acquired by Niantic in 2018. Diana returned to YC in 2021 as a Visiting Group Partner, and joined full-time as a Group Partner in 2022. Over the past four years, she's become one of the most prolific partners in YC's history. She's worked with nearly 230 companies across 18 batches, logged over 2,100 office hours, and those companies are now worth a combined $7 billion.

Diana grew up in Chile and studied computer vision and machine learning at Carnegie Mellon. She built her technical foundation early: data science at OnCue before Verizon bought it, ML research at Intel Labs, then Escher Reality as CTO. At Niantic, she ran the AR Platform team and shipped AR to the 100M+ people playing Pokémon GO.

Few people have both built a startup from zero and shipped tech at that scale. Diana has done both. That combination, paired with deep technical experience in ML and AR, is exactly what today's founders need from a partner. Diana has been at the center of working with the builders pushing the frontier on AI, robotics, and hard tech, including breakouts like Avoca, Reducto, David AI, Salient, Stepful, and HappyRobot.

Congrats, Diana. We're lucky to have you.

We're thrilled to announce that Harshita Arora is joining Y Combinator as a General Partner.

Harshita has been a Visiting Partner at YC since the Summer 2025 batch, making her the youngest visiting partner in YC's history.

Harshita discovered coding at 13, dropped out of school at 15 to pursue it full-time, and by 16 had built Crypto Price Tracker, a crypto portfolio management app that was featured by Apple, which was subsequently acquired, and earned her India's Bal Shakti Puraskar, one of the country's highest honors for young achievers. She then received an O-1 visa, moved to San Francisco, and co-founded AtoB (YC S20).

Harshita and her co-founder came into YC with an idea that got killed by Covid, and with zero background in trucking or payments, they spent weeks visiting truck stops until they found the idea that would become AtoB. Today, AtoB is a Series-C stage company building financial infrastructure for the trucking industry. Often described as "Stripe for Trucking," it offers fleet cards, instant payouts, and modern financial tools, serving over 30,000 fleets across the US.

As a General Partner, Harshita will work directly with YC founders at every stage of their companies. She brings deep fintech and infrastructure experience, a founder's instinct for product, and the perspective of someone who's been building companies since she was a teenager.

We're super excited to have her on the team. Welcome, Harshita!