ML & Agentic Systems Engineer [IC4]
Sourcegraph · Engineering · Lead · Today · Checked today
Open to any country.
Open worldwide with no geographic restrictions, but requires at least 20 hours per week of working-hours overlap with EST.
Direct from the employer's own careers page
Who can apply
Read straight from the listing, so you know before you spend time on it.
Open to any country.
“While we hire almost anywhere in the world, we have a preference for someone to reside in the following locations for this role. However, if you feel qualified, we welcome you to apply regardless of location. No matter what, working hours must overlap with EST for at least 20 hours/week.”
Globally distributed team requiring at least 20 hours per week of overlap with EST.
- Async-first
- The listing doesn't promise async-first working.
- Contractor / B2B
- No mention of hiring international contractors or invoicing.
- Global stipend
- No home office or equipment budget mentioned.
- Annual retreat
- No company-paid get-togethers mentioned.
Read from the listing automatically. Always confirm the details with the company before you accept an offer.
About the role
As the staff ML and agent engineer on the Code Understanding team, you will serve as the technical authority for models, evaluations, and agentic systems powering features like Deep Search and Query Assist. You will design and harden multi-step agent loops, implement rigorous evaluation frameworks, manage model upgrades and fine-tuning, and treat cost and latency as critical product features. Operating on a senior-leaning team, you will tackle ambiguous technical challenges, steer technical strategy, and mentor teammates across production AI engineering.
What you'll do
- Design and harden multi-step, tool-using agent loops to turn research prototypes into scalable, reliable enterprise features.
- Establish pragmatic evaluation suites, smoke tests, and guardrails to reliably detect quality and cost regressions.
- Lead model selection, upgrades, and fine-tuning strategies across Code Understanding surfaces.
- Optimize retrieval, ranking, and context window engineering to ground model outputs in enterprise codebases.
- Profile, distill, cache, and right-size models to optimize cost and latency within strict per-user budgets.
- Participate in the team's on-call support rotation and mentor engineers on agent engineering practices.
What you bring
- Staff-level technical leadership experience building and operating production machine learning and agent systems.
- Demonstrated expertise in agent engineering, including tool use, multi-step execution, and context management.
- Strong background in LLM evaluation methodologies, benchmarking, and observability at scale.
- Proven ability to optimize model latency and inference costs in high-throughput enterprise products.
- High agency and ability to communicate directly with customers to frame and solve complex problems end-to-end.
About Sourcegraph
Sourcegraph builds the context layer and code intelligence platform that helps developers understand, fix, and automate changes across complex codebases. Backed by Sequoia, a16z, and Redpoint, Sourcegraph operates as a globally distributed team serving companies like Stripe and Reddit.
View Sourcegraph profile & open roles →