- Build, optimize, and maintain end-to-end RAG pipelines (document ingestion, chunking, embeddings, vector indexing, retrieval, and reranking).
- Integrate LLM APIs (Anthropic Claude, OpenAI, Gemini, open-source models) into scalable backend services using FastAPI or Python.
- Use AI-native coding setups (Claude Code CLI, Cursor, GitHub Copilot) to explore codebases, write unit tests, and accelerate deployment cycles.
- Design structured outputs, function/tool calling, and agent workflows using frameworks like LangChain, LlamaIndex, or native SDKs.
- Implement basic evals, guardrails, and logging to reduce hallucinations, measure retrieval quality, and monitor API costs.
Who can apply?
Programming: Solid foundation in Python (async, REST APIs, clean object-oriented code) and Git.
GenAI & RAG: Hands-on experience with vector databases (e.g., Chroma, Qdrant, Pinecone, or pgvector) and embedding models.
AI Tooling Native: Daily, active user of CLI or IDE agentic coding tools (Claude Code, Cursor, or similar)—you know how to guide AI agents and rigorously inspect generated code.
Proof of Work: At least 1 shipped or working project beyond basic chat (e.g., semantic search tool, custom RAG on documents, automated agent, or GitHub repo).
Skills you’ll bring
Familiarity with Docker, Linux environment, and basic cloud deployment (AWS/GCP).
Experience with Model Context Protocol (MCP) or multi-agent orchestration frameworks.
Exposure to front-end integration (Streamlit, Next.js, or React basics).