Product Engineer (US)

You've probably been to a therapist, or know someone who has. You also probably know that paying for one is broken in ways most things in 2026 aren't. They "think" they take your insurance. The benefits summary you call to verify turns out to be wrong. Six weeks later, a bill arrives for an amount nobody warned you about.

Behind every one of those moments is a clinic doing the work by hand. Armies of people make eligibility calls, fax prior auths, type claims into payer portals one at a time, and chase denials months after the visit. Good clinicians leave the field over it.

Supa Health is building the agentic operating system that makes this work disappear — an AI scribe, an AI receptionist, and an agentic back office that handles benefits verification, prior auth, intake, and claims. Every call its agents make, every denial they work, every payer quirk they encounter feeds back into a system that gets sharper the next time. We're building software that compounds.

We're not selling dashboards. We're replacing the work itself.

Small team, $5M from Peak XV (formerly Sequoia India), shipping fast. Looking for our founding product engineer.

The honest answer is: whatever moves the product furthest. We have surface area in every direction and we'd rather you pick the one that fits you than wedge you into a slot.

That could mean:

• Operator-facing surfaces. Real-time, dense, information-rich SPAs where our back-office team works through hundreds of cases a day. The interaction model matters as much as the engineering — operators live in these tools eight hours a day, and small wins compound into hours saved. Modern React stack, TypeScript end-to-end, real-time over WebSockets.

• The agent platform. The systems that let our agents do real work in the world — making calls, navigating systems built before APIs existed, parsing structured and unstructured payer responses, and writing back into our own state. Worklfows for orchestration, Postgres as the source of truth, LLMs invoked deliberately rather than reflexively.

• The platform underneath both. Schema, services, auth, multi-tenancy, integrations, and the HIPAA-grade segregation that has to be right on day one. Postgres, Drizzle, Fastify, an Nx monorepo, WorkOS for auth.

• The next product wedge. We have a roadmap. We also have customers asking for things that aren't on it. Founding engineers at Supa Health get to argue for what we should build next, and then build it.

You'll talk to operators. You'll talk to clinic owners. You'll watch someone do their job for two hours and then go automate the painful parts of it. You'll ship something on Tuesday and see it running in production on Wednesday.

• Craft. You've shipped real software to real users for years and it shows in how you think. You have opinions about schema design, about when to reach for a state machine, about what makes a UI feel fast. You can tell the difference between extra effort and meaningful quality, and you reach for the second one. Your code reviews make the people around you better — not because you're harsh, but because you're specific, and the team's standards rise around you.

• Judgment. Healthcare is full of edge cases that look like bugs and bugs that look like edge cases. The CTO can't be in every decision and we don't want to use process as a substitute for thinking. We want people who can reason through a tradeoff with incomplete information and make the call.

• Ownership. You don't wait for work to be handed to you. You see what's broken, what's slow, what's missing — and you write the doc, file the issue, or open the PR. You take a problem from "I noticed this" to "it's deployed and working," which means talking to the user, designing the solution, writing the code, watching it land, and going back to fix what you got wrong.

• Direction. You work with users and teammates to figure out what to build and how to build it. You sit with operators, watch where they get stuck, and come back with a sharper definition of the problem than we started with. You're a thought partner to the team — not someone waiting for a spec, and not someone going off to build in a vacuum.

• Speed. Not crashed-out hustle-culture speed. The kind of speed that comes from deleting unnecessary work, making decisions quickly, and not being precious about being wrong.

• Range. You don't need to be a 10/10 at frontend, backend, infra, AI, and product. You do need to be willing to learn whichever of those the work demands this week.

5+ years building production software. We care less about the years and more about the slope — what you've shipped, what you learned the hard way, and what you'd do differently now. You've lived in TypeScript and Postgres long enough to have real opinions about both.

Bonus, not required:

• Agentic AI / LLM systems. You've built something with agents, tools, evals, and the unglamorous plumbing around them. You know when an LLM is the right answer and when it's a fancy way to make a slow, wrong API call.

• Early-stage startup experience. You know what it feels like to have no design system, no playbook, no PM, and a customer call in 90 minutes.

• Healthcare / RCM background. You know what a 271 is. You've fought with a clearinghouse. You have feelings about Availity. This is rare and we'd love to find it, but we'll happily teach the right person.

• Compensation. Salary in the $190K–$230K range plus meaningful founding-team equity, both calibrated to seniority and scope. We'll be specific in the first conversation.

• Health, dental, vision. Fully covered.

• 401(k) with match.

• In-person in the Bay Area is the default — founding teams move faster in a room together. We'll consider remote or hybrid setups for exceptional candidates. Commuter benefits for those in office.

• Meals. Lunch, dinner, and snacks on the company.

We borrowed our process from Linear because we think they got it right.

1. Intro call with the CTO. ~30 minutes. We talk about you, the company, and whether the role fits. We'll discuss comp on this call.

2. Technical conversation. ~60 minutes. We work through a real problem together. No leetcode.

3. Paid work trial. 2–5 days, paid at our standard daily rate. You work on something real. You meet the team. You leave with a much clearer picture of whether Supa Health is the right place for you, and we leave with a much clearer picture of whether you're the right person for Supa Health.

4. Decision. Independent feedback from the trial team, then a debrief, then an offer (or honest feedback if not).

We aim to go from intro call to offer in 1–2 weeks.

We'd rather be honest now than waste your time later.

• If you need a roadmap handed to you, you'll be uncomfortable here. The roadmap is a conversation, and you're in it.

• If you want a 9-to-5, that's a legitimate and healthy choice, and it isn't this job. We don't expect crunch, but we expect founding-team intensity.

• If you're allergic to talking to users, this won't work. Operators and clinic owners are part of your job, not someone else's.

• If you want to manage, also not this. Founding engineers here build. People-management roles will exist later; this isn't one.

If you read all of that and you're more interested than when you started, we want to talk.