Core R&D Engineer - Agentic Systems & Retrieval

Hybrid (India / UAE) · Full-time · 4–10 yearsWe build agentic AI systems for institutional investors. This role is on the platform team: you build the orchestration, retrieval and evaluation layers that everything else runs on.What you'll doDesign and ship multi-agent workflows in LangGraph, LangChain and Google ADK — routing, delegation, durable and resumable execution, human-in-the-loop checkpointsBuild our hybrid retrieval engine across three modes: knowledge graph, vector search, and SQL over structured data — plus the router that decides which to use and fuses the resultsWork across Gemini, OpenAI, Azure OpenAI and Anthropic models, with model-agnostic routing and fallbackOwn the agent harness: tools, MCP integrations, guardrails, context and token budgeting, tracingBuild evaluation infrastructure — golden datasets, regression suites, grounding and hallucination checksShip production Python and Node.js/TypeScript services, with React front-ends where neededWhat we're looking for4+ years shipping production software; strong in both Python and Node.js/TypeScriptReal production experience with an agent framework (LangGraph, LangChain, ADK or similar) — not tutorialsYou've built a RAG system and then fixed it. You can talk about retrieval failure modes, chunking, reranking, and what didn't workStrong SQL; graph databases (Neo4j/Cypher) or the ability to pick them up fastDocker, Kubernetes, CI/CD — you deploy what you buildYou write evals. You've caught a regression before a user didBonus: knowledge graphs and entity resolution · text-to-SQL · open-weight model serving (vLLM) · Firestore/Postgres at scale · open-source contributions