Tiny Health - AI Engineer
Requirements
• 4+ years building and shipping software, including hands-on experience taking ML or LLM-backed features to production. You can walk us through what broke once real users depended on it, and what you did about it. • Solid machine learning fundamentals. You understand how the models you work with actually behave and can reason about their limits, rather than treating them as black boxes. • Working knowledge of retrieval-augmented generation: embeddings, chunking, retrieval quality, and the reasons a retrieval system returns the wrong thing. • Strong backend engineering fundamentals, with experience in TypeScript or Python. • Ability to debug non-deterministic systems, isolating whether a bad output came from retrieval, prompting, tool use, or post-processing. • Treats evaluation as part of building, with experience measuring AI quality rather than relying on spot checks. • Experience with AWS and serverless architectures. • Strong understanding of API development and system design. • Comfortable driving work forward in a fast-paced, remote startup environment. • Preferred • Familiarity with vector databases (Pinecone, Weaviate, FAISS, Chroma, or similar). • Experience with LLM-as-judge evaluation or human-in-the-loop feedback loops. • Experience with agent tooling, function calling, or MCP. • Experience with observability, monitoring, and alerting systems. • Previous experience in healthtech, biotech, or handling sensitive health data. • Experience fine-tuning and self-hosting models for narrow, latency-sensitive tasks. • Startup or small-team experience where you've worn multiple hats. • How we work • How we work • A mission that actually pulls people inWe make the microbiome measurable and actionable at every life stage, and real people write in to say it's life-changing. • A mission that actually pulls people in • Remote-first, real overlapOur FTEs are based in North America and work core hours from 9am–6pm CST, shiftable by up to two hours either way — start at 7am or 11am, wrap at 4pm or 8pm. • Remote-first, real overlap • Written first, huddle second, meet lastDefault to a clear Slack post, huddle when a thread stalls, and meet only to make a real decision. • Written first, huddle second, meet last • Fast, but never sloppyDo the two-month thing in two weeks, then check the work—shipping errors is rework, not speed. • Fast, but never sloppy • Shape what we build, not just how we build itThere's no playbook waiting for you, so write it—and bring the alternative, not just the objection. • Shape what we build, not just how we build it • Day One mentalityFewer titles and layers; process is a guardrail where it counts, never the reason something didn't ship. • Day One mentality • No hidden agendasNumbers, misses, and feedback are shared openly and in real time, so you always know where you stand. • No hidden agendas • High standards, genuinely fun peopleTake the work seriously and ourselves much less—high-performing and fun to be around is the hiring bar. • High standards, genuinely fun people • Is this pace for you?Venture-scaled and on a steep trajectory, with intensity in spikes, fiercely protected flexibility, and faster growth than anywhere else. • Is this pace for you? • Our Values • Our Values • Learn Fast, Get Better - Find the root cause, not just the fix, and get better every time • Learn Fast, Get Better • Be Relentlessly Resourceful - Dig for the answer, move fast, and know when to ask for help • Be Relentlessly Resourceful • Act Like an Owner, Be Hungry to Win - Act like it’s your company, and do whatever it takes to win • Act Like an Owner, Be Hungry to Win • Act with Honesty and Empathy - Say what’s true, and say it with care • Act with Honesty and Empathy • Delight People by Anticipating Their Needs - Solve the problem, then get two steps ahead of it • Delight People by Anticipating Their Needs
Responsibilities
• Design, build, and maintain production AI systems and APIs that power our products. • Own AI features end to end, from prototyping and evaluation through deployment, monitoring, and iteration. • Build and improve retrieval and grounding systems so our answers stay accurate and traceable to their sources. • Develop the evaluation and quality tooling that tells us whether a change made an AI system better or worse. • Ensure the performance, reliability, and observability of AI components in production, including triaging and resolving quality issues as they surface. • Collaborate with our science and clinical teams to translate reviewed microbiome science into product experiences people can trust. • Partner with product, design, and engineering to scope and ship high-impact AI-driven features. • Design data pipelines and inference infrastructure that scale as our usage grows.
Apply in one click
Upload My Resume
Drop here or click to browse · Tap to choose · PDF, DOCX, DOC, RTF, TXT