Applied AI Engineer

Amigo•San Francisco, CA

5d•$130,000 - $240,000•Onsite

About The Position

Applied AI is how we turn a customer's hardest clinical workflow into production AI that actually ships. As an Applied AI Engineer, you own the technical delivery of real deployments end to end. You design the agents, build them, and are on the hook for whether they work in the field. The team is small and dense, and you get a lot of surface area early. We hire Applied AI Engineers across every level, from new grads to deeply experienced engineers, and we work out level together once we've met you. What matters is not years on a resume but whether you can take an ambiguous problem and turn it into something reliable that patients and clinicians depend on.

Requirements

You've built real software that other people relied on, and you care whether it actually works, not just whether it shipped
You're comfortable with LLMs and agents, or you'll get comfortable fast
You take ownership. When something is broken you fix it, and you say so early when something is more than you can carry
You have good judgment about tradeoffs, and you'd rather get the direction right than move fast in the wrong one
You're low ego, direct, and hold yourself to a high bar
You can work on site in San Francisco
This role spans new grads through senior and staff engineers. Strong early-career people are welcome, and so are people who have done this work for a long time.

Nice To Haves

Experience in a regulated or high-reliability domain (healthcare, finance, legal)
Background with evaluation, simulation, or synthetic data
Experience working directly with customers or domain experts
Python depth and experience building reliable systems on top of external APIs

Responsibilities

Designing and building an AI agent for a specific clinical workflow, from first prototype to production
Deciding what an agent should and should not handle, and where it hands off to a human
Building the evaluations that decide whether an agent is safe to ship, and catching failure modes before patients do
Writing the integrations that let agents work inside a customer's existing systems
Debugging why a live conversation went wrong and fixing it across the whole stack
Finding the patterns that repeat across customers and turning them into something reusable