About the role
We're growing our Python AI Engineering team - the group that builds GenAI systems clients actually run, not proof-of-concepts that die after the demo. Expect real agentic/RAG systems, real users, and evaluation data that tells you whether what you built actually works.
Depending on fit and timing, you'll be matched to one of several active or upcoming client engagements - each with its own domain, constraints, and stack, but the same bar for engineering quality.
You are a good fit for the role if...
- You default to shipping, not just designing - a proactive, "getting things done" attitude in a fast-paced, client-facing setup where the interesting problems rarely come pre-scoped in a ticket.
- You own the AI lifecycle end to end - ideation, architecture, delivery, evaluation - and you're comfortable being the technical voice in the room: proposing the approach, defending it with data (evals, latency, cost), and adjusting it when the data says you're wrong.
- You can translate engineering constraints into terms a client stakeholder acts on - not just "this won't scale" but why, what the tradeoff actually costs them, and what a workable alternative looks like. Good technical judgment shows up as much in how you explain a constraint as in identifying it.
- You treat client pushback as a design input, not an obstacle - sometimes it reveals a real business constraint you were missing, sometimes it's a shortcut worth resisting. Telling those two cases apart, and being able to argue either side with evidence, is the actual skill.
Our expectations
- 5+ years commercial Python experience, including 3+ years of hands-on GenAI/LLM engineering.
- Solid engineering fundamentals: building scalable services/APIs (FastAPI, Flask, or Django), writing real test suites (pytest), clean and modular architecture.
- Hands-on experience building RAG pipelines - retrieval, embeddings, vector search.
- Working knowledge of agentic patterns: tool-calling, function-calling, multi-step reasoning workflows.
- Strong prompt engineering skills, including structured outputs (JSON schemas, Pydantic, Instructor or equivalent).
- Experience with at least one major cloud platform (AWS, Azure, or GCP), including its managed AI/ML services (e.g., Bedrock, Azure OpenAI).
- An "evals mindset" - you think about relevance, consistency, latency, and cost as real engineering concerns, not afterthoughts.
- High-proficiency written and spoken English - you'll use it daily with clients and teammates.
Nice to have
- Experience with orchestration frameworks beyond the basics - LangGraph, LangSmith, LlamaIndex.
- Hands-on with a specific vector database (Pinecone, Weaviate, Milvus, pgvector) beyond "I integrated one once."
- Experience building evaluation frameworks or golden-dataset pipelines specifically (as opposed to just using one).
- Exposure to data pipeline work feeding AI systems - understanding how data quality/freshness affects model behavior.
- Prior client-facing / consulting experience in a professional-services or consulting setup.
Main responsibilities
- Design and build Python services and APIs that wrap LLM-powered functionality.
- Build and maintain agentic and RAG pipelines: retrieval, reranking, tool-calling, multi-step reasoning, structured outputs.
- Care about quality beyond "it works" - evals, observability, and the data that tells you when something regresses.
- Work directly with clients: translate fuzzy business requirements into architecture decisions, and explain your technical tradeoffs to non-technical stakeholders.
- Work across a distributed, multi-market team and client organizations.
