About the job
We're building the AI-powered backbone of our developer tools. You'll work on three core systems: a retrieval infrastructure powering our AI products, a coding agent that assists users in writing code, and the evaluation frameworks that measure their effectiveness.
Responsibilities
- Design and scale retrieval pipelines including vector search, BM25, and hybrid retrieval strategies
- Build and optimize embedding pipelines, chunking strategies, and re-ranking systems
- Develop query understanding and rewriting components to improve retrieval relevance
- Manage vector database infrastructure at production scale
- Build agent architectures that support multi-step code generation, refactoring, debugging, and diagnostics
- Implement tool-use patterns (function calling, code execution sandboxing, file system interaction)
- Develop context management strategies for long-form code understanding
- Design and maintain evaluation frameworks for retrieval quality and agent task completion
- Build custom benchmark suites for real-world coding task assessment
- Create reproducible testing infrastructure with automated regression detection
Qualifications
Minimum
- Bachelor's degree in CS, EE, or related field (or equivalent experience)
- 4+ years of software engineering experience
- Strong proficiency in Python (primary) and TypeScript/JavaScript
- Experience with at least one systems-level language (C++, Rust, Go)
- Practical experience integrating LLMs into applications (prompt engineering, context management, output parsing)
- Understanding of agent patterns: tool use, multi-turn reasoning, error recovery
- Familiarity with code-specific LLM tasks (generation, summarization, analysis)
Preferred
- Master's or Ph.D. in Computer Science, Machine Learning, or related field
- Contributions to open-source developer tooling, retrieval systems, or coding assistants
- Experience with AST parsing, code analysis tools, or language servers
- Familiarity with rendering pipelines or cross-platform framework architecture
- Experience deploying and optimizing ML models in production (latency, cost, reliability)