Sourcegraph
ML & Agentic Systems Engineer [IC4]
About this role
Sourcegraph seeks a Staff ML and Agentic Systems Engineer to lead the technical direction of their Code Understanding team, owning production ML systems, agent engineering, and evaluations that power deep code search and AI-assisted development tools. You'll combine software engineering, machine learning, and statistics to make their agentic products measurably better, faster, and cheaper at enterprise scale.
What you'll do
- Design and harden multi-step, tool-using agent loops for reliable, observable, and cost-effective agentic experiences
- Establish evaluation strategies and metrics that balance rigor with speed, guiding when to use smoke tests vs. comprehensive evaluations
- Own model selection, upgrades, and fine-tuning decisions to optimize for performance and cost
- Engineer retrieval and context systems to ground models in customer code with improved accuracy and verifiability
- Profile and optimize cost and latency as core product features through distillation, caching, and model right-sizing
- Set technical direction and standards for the team's work with models and agentic systems
What they're looking for
- Production machine learning systems and MLOps
- Agent engineering and multi-step reasoning systems
- LLM fine-tuning, distillation, and model evaluation
- Retrieval-augmented generation (RAG) and context engineering
- Evaluation design and metrics for ambiguous, evolving systems
- Cost and latency optimization at scale
- Python or similar production ML languages
- Software engineering and system design
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Sourcegraph
Sourcegraph builds a code intelligence platform that helps developers understand and navigate large codebases through AI-powered features and open-source indexing protocols. The company is hiring for infrastructure engineers, AI/agent specialists, compiler engineers, security professionals, and customer-facing field engineers to scale its SaaS offering and enterprise deployments.
- Website
- sourcegraph.com
Likely interview questions
- Walk us through how you've designed and debugged a multi-step agent system in production—what were the hardest parts?
- How do you approach evaluation when the system and product are both changing rapidly? Give a concrete example.