Role Summary
The technical owner of the Core Intelligence layer — the platform’s differentiator. You design and build the Signal Engine, Action Engine, Librarian Agent and Ontology Graph, turning signals into coordinated, agent-driven action.
Key Responsibilities
- Design and implement agentic workflows with LangGraph and Azure AI Foundry that power the Signal and Action Engines.
- Build the Librarian Agent — retrieval-augmented reasoning over the Ontology Graph and Graph Database.
- Model and evolve the Ontology Graph; implement graph and vector retrieval strategies.
- Integrate agents with tools and other services via MCP / A2A and webhooks.
- Establish evaluation, guardrails, observability and cost controls for LLM/agent behavior in production.
- Partner with the Data Engineer on the data and features that feed the intelligence layer.
Must-have Skills & Technologies
- Strong Python and production ML/AI engineering.
- Agentic frameworks — LangGraph (or LangChain, LlamaIndex, Semantic Kernel) — and LLM orchestration patterns.
- Azure AI Foundry or equivalent (Azure OpenAI, Bedrock, Vertex) model deployment and management.
- RAG, embeddings, vector search, and graph-based retrieval; hands-on with a Graph Database (Neo4j, Cosmos DB Gremlin or similar).
- Prompt engineering, structured output / tool-calling, and LLM evaluation frameworks.
Prior Experience
- 5+ years in ML/AI engineering, including production LLM or agent systems (not only notebooks/POCs).
- Experience taking an agentic or RAG system from prototype to reliable, monitored production.
Nice to Have
- Ontology / knowledge-graph modelling; MCP / A2A protocol experience.
- MLOps / LLMOps: eval harnesses, prompt versioning, drift and cost monitoring.
