AI Engineer

AI features inside products people use daily, not proofs of concept. On one healthcare marketplace we built a conversational layer over SMS, email and chat that qualifies candidates, collects documents, and hands off to a human recruiter mid-conversation, plus a scoring model that tells recruiters who to call first.

Work in this seat looks like: designing the handoff so it doesn't feel like a bot gave up; building the evaluation set that catches a regression before the client does; deciding which three signals a score actually needs when eight were specified; and running the cost and latency numbers before anyone commits to a model.

• 2+ years of applied AI work in products that shipped, not only notebooks or demos
• Strong Python; comfortable with at least one model provider API and the tradeoffs between them
• Retrieval done properly: chunking, embeddings, reranking, and knowing when RAG is the wrong answer
• Prompt and system design you can explain and defend, with evaluations to back it
• You measure quality rather than guessing at it

• Agentic workflows and tool use in production
• Fine-tuning, distillation, or running open models where cost demands it
• Vector databases at scale, or a good argument for not using one
• Experience with compliance-heavy domains like healthcare