AI Product Engineer - Series B Healthcare AI (SoMa, San Francisco)
San Francisco (SoMa) · 4 days onsite · $180K–$285K base + competitive equity · Visa sponsorship considered (H-1B transfers, TN, OPT) · Relocation support available · 2–4 hires
The short version
I'm running an exclusive search for a Series B healthcare AI company in San Francisco. Series B just closed with two of the best-known Sand Hill Road names on the round - general partners themselves, not associates. Revenue went from $200K to $10M last year . Product is live inside two of the highest-bar academic medical centres in the US. Named World's Most Innovative in 2026 and Best in AI Implementation in 2025.
They are 24 people, 7 in engineering, deliberately keeping the team small. The seat is an AI Product Engineer working directly with the VP of Engineering on the platform that lets health systems build, share and deploy AI agents for clinical and operational work.
Around fifty patients so far have had treatment they otherwise wouldn't - and the reason each time is a workflow one of these engineers shipped. That's the tone of the work.
What they're building
A platform for health systems to build, share and deploy AI agents across clinical and operational tasks. The mental model is the three-step loop that runs most clinical work: pull the knowledge, reason over it, take the action. Same product shape as the agent platforms hitting sales and support right now, in a much harder industry - the users aren't developers, the stakes are life-ending and the buyers are among the most conservative in the world.
They bet early on agentic capability. That means every frontier-model release makes the product materially better out of the box and the engineering team gets to focus on the layer above - orchestration, evals, primitives, the last-mile UI a clinician actually clicks.
What you'd actually do
- Own product features end-to-end - ideation, design, build, ship, deploy. There are no PMs.
- Build and scale agentic workflows and agent primitives for reliable long-running clinical reasoning.
- Roughly 65% backend, 35% frontend on the first hire. Python for backend, React/TypeScript for frontend. Rust and Kubernetes are learnable on the job - don't rule yourself out on those.
- Split your time roughly half on code and half on planning, reviewing, and thinking about user impact and architecture.
- Interface directly with clinicians, end-users and customer call recordings. You are the one hearing the pain and shipping against it.
- Contribute ~60–70% on top of existing infrastructure, ~30–40% net-new.
The team is flat, async and product-driven. No standups. No PMs. Everyone reports to the VP of Engineering. If you like sprints, tickets and layered management, this is not for you!
Who they're looking for
2-8 years of full-time software engineering - sweet spot 4-6. The seat is squarely IC-builder; engineers who've moved into management or want to, this isn't the one.
CS degree is stated. Strong Python for backend is non-negotiable. React/TypeScript on the frontend. Familiarity with Rust, Kubernetes, Helm and AWS is welcome but the team have said clearly these are learnable on the job.
The load-bearing gate (and the one worth being honest with yourself about) is what the HM calls "one clear peak." At least one of the following on your CV:
- Strong CS degree (think top-10-adjacent - Stanford/MIT/Berkeley/CMU tier, but the team has interviewed strong Berkeley and UCLA profiles, so read this as "top-tier practical bar" rather than a strict ranking).
- Fast promotion at a recognisable engineering-culture company.
- Founding engineer at a VC-backed startup during a real growth phase.
- Standout team at a hard-to-enter organisation (Stripe/Palantir/Scale/Databricks/Anthropic-tier).
Zero peaks is an auto-reject. One clear peak is the floor. Two is comfortable.
Beyond the peak gate, what actually clears interviews:
- Real end-to-end product feature ownership without a PM holding the pen.
- Genuine production experience with agentic workflows or LLM-based systems - shipped, not "familiar with."
- Product thinking. Because there are no PMs, the interview loop specifically tests whether you can reason about end-user impact, not just implementation.
- High ownership. The team will not chase you.
- Bonus, not required: healthcare or healthtech background.
What "great" looks like (in the HM's own vocabulary)
There's a distinction worth knowing up front, because it's the lever on the pay band:
- Solid - clears the peak gate, executes independently, owns a vertical, learns the domain on the job. Lands mid-band.
- Great - multiple peaks on the CV (top school + name-brand company + founding shape at a fast-growth startup), directly relevant agentic/production AI experience, product intuition already there. This is the profile that unlocks the top of the band and full flexibility on relocation.
The ideal-profile calibration the HM shared makes this concrete. The top of the list is a top-10 CS graduate who ran forward-deployed at a name-brand data-platform company and then went early at a healthtech scale-up. The would-interview cutoff sits at "school not top-10 but founding AI engineer at a real Series B" - one strong marker, clean shape.
What won't work
Straight from the client, so you don't waste your time or theirs:
- Zero peaks on the CV - no strong school, no name-brand company, no founding-shape at a growth startup, no standout-team at a hard-to-enter org. This is the wall.
- Legacy enterprise or IT-services backgrounds - Oracle, SAP, IBM, Infosys, Wipro, Cognizant, Accenture, Capgemini-shape. Multiple candidates from these backgrounds have struggled in the interview loop and the HM has told me directly it's a heuristic reject.
- Long careers in Java-on-Spring, big teams, long planning cycles. This environment burns that shape and they know it.
- Serial job-hopping - multiple founding roles under one year each without a real story is a stated trait-to-avoid.
- Frontend-primary CVs. Role is 65% backend, so a UI-only career is a fit-shape reject even if the peak is there.
- Senior engineers who've moved into management and want to stay there. The seat is IC.
- Currently on your final H-1B lottery attempt with no overseas fallback. Small US-only team, nowhere to move you to if the lottery doesn't come through.
The details
- Location. San Francisco, SoMa. 4 days a week onsite. Relocation support is genuinely on the table for the right candidate.
- Comp. $180K–$285K base plus competitive equity. Strong Series B numbers, not frontier-lab numbers.
- Visa. Open to new H-1B, H-1B transfers, TN, OPT. Preference caveat above on final-lottery-attempt profiles.
- Team. 24 people total, 7 in engineering. Flat, async, no standups, no PMs. Reports into the VP of Engineering.
- Timeline. First hire in 3-4 weeks. Subsequent hires roughly monthly.
Why this one is genuinely worth a conversation
- The investors and the traction are real. $200K to $10M in a year, live inside two of the highest-bar academic medical centres in the country, Series B closed with GPs personally on the round. Different pressure profile from most Series Bs - they're not raising in a hurry.
- The engineering shape is the actual selling point. Flat 7-person team, no PMs, no standups, everyone owns product end-to-end. If you've been the engineer quietly holding the product together without the org chart to match, this is a seat where that's the job description, not the workaround.
- The AI work is the product. Agent primitives, orchestration, evals on live clinical workflows - not a bolt-on. Every frontier-model release makes the product better; the team gets to focus on the layer above.
- Foundational hire. Going from 7 to 9–11 engineers over the next six months, then holding lean for a long time. The patterns and tooling you set will be what everyone builds on afterwards.
- The domain is unglamorous and important. Fifty patients so far have had treatment they otherwise wouldn't have. That number will grow and the reason each time will be a workflow an engineer on this team shipped.
Interested?
Apply and lets see if there's a fit - 3 things I'll want to cover on our first conversation:
- The one clear peak on your CV - school, company, founding-shape, or standout-team - and the shape of your career around it.
- A recent agentic or LLM system you actually shipped to production - what did you build, and what did the evals and guardrails look like?
- Confirmation you can be onsite in SoMa four days a week (or a real conversation about relocation, if that's the ask).
If it's a fit, I'll walk you through the client, the team and the process in more detail!
Thanks,
Tom Calver · HorizonAI Talent · tom@wearehorizon.ai