Join a fast‑moving team building a coding agent that autonomously modifies customer codebases, verifies changes, and operates securely within on‑prem environments. You’ll bridge AI research and production tooling to deliver safe, interpretable software automation.
What You’ll Do
- Design agent architecture and tool interfaces for code retrieval, editing, builds, and tests.
- Implement verification pipelines that ensure changes are correct before any human review.
- Build long‑horizon execution with checkpointing, recovery, and self‑correction for multi‑file refactors.
- Define authorization and rollback models governing agent actions and undo mechanisms.
- Integrate observability tooling for debugging, audit trails, and performance monitoring.
What You Need
- Deep experience building IDE extensions, language servers, or other developer tooling.
- Expertise in code intelligence: ASTs, static analysis, symbol indexing, and automated refactoring.
- Proven work with build systems and CI/CD pipelines (Bazel, Maven, GitHub Actions).
- Strong background in observability: OpenTelemetry, distributed tracing, and structured logging.
- Advanced Python development for backend services, sandboxing, containerisation, and cloud/on‑prem infrastructure.
Good to Have
- Experience building evaluation datasets or frameworks for AI systems.
- Familiarity with agentic frameworks (LangChain, LangGraph, Semantic Kernel) in production.
The Opportunity
Lexsi Labs has published over 25 papers in the past 15 months at top conferences, and engineers work in a flat, high‑autonomy environment where ownership spans from architecture to production.