Job Description

We're building production AI agents that change how our engineering org ships code and how employees work with internal knowledge. You'll join a small, high-ownership team. You'll write the agent logic, the tools, the data layer, and the pipelines that make them dependable at scale. Turn architecture into reliable, well-tested, production-grade systems.

Role and Responsibilities :

• Implement agent logic, tools, and orchestration against the given architecture.
• Build and tune RAG pipelines ingestion → embeddings → retrieval → re-ranking → evaluation
• Design and optimize Postgres schemas, queries, and vector indexes for production load
• Build the test-generation and test-execution engine for the AI Code Pipeline
• Integrate with Git, CI/CD, and internal services
• Build evaluation harnesses, observability, and guardrails so agents are measurable and safe
• Own the reliability, latency, and cost of the systems you ship

• Write clean, typed, well-tested Python and participate actively in code review

Requirements

• 4+ years of professional software engineering in Python, with deep, hands-on command of the language : async, typing, packaging, performance, and clean, maintainable code
• 1+ years building LLM / agentic systems — tool use / function calling, orchestration, prompting, and evaluation
• Strong problem-solving and debugging on ambiguous, real-world production problems.
• Solid Postgres experience: schema design, query optimization, and extensions ( pgvector)
• Experience shipping and operating production services — testing, CI/CD, observability, on-call mindset
• Comfort with Git-based workflows and code review
• Automating software testing: unit, integration, functional, security, performance

• Integrating with developer tooling and CI/CD systems

• 6+ months of demonstrated experience building with Claude Code in a development workflow

Benefits

Opportunity to work for a dynamic international company with a flat hierarchical structure, where your voice matters and your impact is seen.