Job Description
Job Description
About this role
We are looking for a Product Engineer (Search) with 3–12 years of experience to own our client's developer-facing search endpoint — closing the gap between retrieval and ranking research and what developers actually feel in the API. You'll be the single person who takes a ranking improvement from research and ships it as a live, polished developer experience in days, not sprints. No PM, no hand-holding — just you, the product, and a fast-moving team building infrastructure for the AI era.
What will you be doing?
-
Own the search API end-to-end — response format, latency, error handling, filtering, and docs — as the sole accountable person for how it feels to developers.
-
Translate retrieval and ranking research wins into shipped product changes developers notice (think: 200ms latency improvements, better recall/precision tradeoffs).
-
Dogfood the API relentlessly — read every GitHub issue and Discord thread touching search and fix friction before users have to ask.
-
Run fast product experiments: form a hypothesis, instrument it, ship it, measure it, and decide quickly with imperfect data.
-
Define what good looks like independently — no PM to scope tickets; you set the priorities and own the outcomes.
Key Requirements
-
Production search experience (retrieval, ranking, relevance) at a developer-facing company — this is a hard requirement, not a nice-to-have.
-
Has shipped on top of Elasticsearch, OpenSearch, Vespa, Algolia, vector databases, or a custom IR stack, and owned an externally consumed search API.
-
Can talk fluently about BM25 vs. semantic hybrid retrieval, re-ranking, and recall vs. precision tradeoffs — unprompted.
-
Founding engineer DNA: has shipped features end-to-end at a sub-200-person startup (Series A–C) without a PM, designer, or QA layer.
-
TypeScript/Node.js proficiency; comfortable with observability tooling (Datadog, Sentry, OpenTelemetry) and cloud infra in production.