ML & Agentic Systems Engineer [IC4] (Remote, Hires Almost Anywhere in the World)

Harden multi-step agent loops that plan, call tools and recover from failure in production. Own the lifecycle of the models behind Sourcegraph features, including selection, upgrades, fine-tuning and monitoring. Build evaluation harnesses and set evaluation standards so the team can measure quality, cost and latency with rigor. Design retrieval and context-engineering approaches that ground LLMs in customer codebases. Drive cost and latency improvements while keeping enterprise reliability. Mentor other engineers and operate autonomously on ambiguous, high-impact problems.

Staff-level technical leader with hands-on production ML and agent-systems experience. Has personally owned a production model lifecycle from training through monitoring and has built multi-step agentic systems. Strong evaluation judgment and the ability to work independently on ambiguous problems. Solid software engineering fundamentals; Sourcegraph works in Go, TypeScript, GraphQL and Postgres. Working hours must overlap with US Eastern Time for at least 20 hours per week. Nice to have: shipped LLM or agentic products, fine-tuned models for production targets, retrieval and ranking experience, enterprise customer work, and mentorship experience.