Software Engineer, Senior

• Own production LLM integration at maturity: structured, schema-validated output enforcement, drift detection, sanity bounds, and the cost and latency guardrails that keep in-product AI features accurate and trustworthy at scale.
• Treat prompts as engineering artifacts — versioned, compared head-to-head, and protected by regression evaluations that gate changes before they ship.
• Design, build, and tune retrieval-augmented generation: embeddings, vector search, retrieval ranking, and grounding/citation enforcement so answers are accurate and never hallucinated.
• Build and operate the upstream content pipeline that authors and maintains the product's use-case catalog — scheduled, reproducible runs with viability scoring, a change-review workflow, and reliable write-back integration into the product.
• Partner with the product engineers to ship in-product LLM features — recommendation generation, semantic search, and seller-prep — with the right accuracy, cost, and latency profile.
• Own the quality and freshness of the catalog and the AI features so they do not go stale or degrade as the product scales, and explain model trade-offs in plain language to product and the field.

• Senior level: ~5+ years building production software, with ~4+ years specifically in applied ML or LLM/GenAI systems engineering. You own work end to end — from an ambiguous problem to shipped in front of real users — and stay responsible for it in production.
• Deep, hands-on production experience building LLM-powered application features against any major model provider or open-weight model — we are provider-agnostic and screen on the capability, not the vendor.
• Retrieval-augmented generation (RAG) as a discipline you can speak to in your own words and have built: embeddings, vector search, retrieval ranking, and grounding/citation enforcement.
• Evaluation and guardrails treated as engineering, not an afterthought: prompts kept as versioned, tested artifacts with eval/regression checks (any framework or your own harness), plus structured/validated output enforcement.
• The production hardening that keeps AI accurate and trustworthy at scale: cost/latency optimization (caching, batching, prompt distillation), drift detection, and sanity bounds.
• Strong in Python OR TypeScript — either qualifies; we are not requiring both.

What We're Looking For

• Has built and launched real products from start to finish. They can clearly describe what they built, who used it, and what they personally did.
• Uses AI coding tools every day to work faster, and can explain how they use them.
• Learns new things quickly. They can give an example of a tool or skill they picked up recently and used well.
• Works well on their own. They can take an unclear problem and move it forward without waiting for detailed instructions.
• Explains technical things clearly to non-technical people — here, engineers work directly with the product team and the sales field.
• Is reliable: follows through without being reminded, raises problems early, and takes responsibility for the result, not just the task.