GenAI Lead Engineer

We are looking for a GenAI/LLM Engineer with strong Python engineering skills and proven experience building production-grade Retrieval-Augmented Generation (RAG) systems using LlamaIndex and/or LangChain, and integrating vector databases (Pinecone preferred). The role will focus on designing scalable RAG pipelines, implementing advanced retrieval strategies, and building MCP/tool-calling connectors to expose enterprise APIs as agent tools for read/write operations.

Key Responsibilities

1) RAG Pipeline Design & Development

• Design and develop end-to-end RAG pipelines, including:
  • Data ingestion
  • Document parsing and preprocessing
  • Chunking strategies
  • Embedding generation
  • Indexing into Pinecone (preferred)
  • Retrieval and response generation
• Build production-ready semantic retrieval solutions and continuously improve relevance/grounding quality.
• Implement and optimize advanced retrieval strategies, including semantic search and retrieval tuning.

2) Agent Tooling & MCP Integrations

• Build and integrate MCP connectors to expose internal/external system APIs as agent-callable tools (read/write).
• Contribute to agent orchestration patterns including:
  • Intent routing (e.g., deciding between RAG vs MCP vs workflow)
  • Tool selection and execution sequencing
  • Agent reliability patterns (fallbacks, retries, observability)

3) Security, Reliability & Performance

• Apply security controls and handle authentication/authorization tokens, ensuring safe access to enterprise systems.
• Optimize AI/ML workflows for performance, scalability, and reliability (latency, throughput, cost, robustness).
• Ensure seamless deployment and integration across environments in collaboration with platform/DevOps teams.

4) Cross-functional Collaboration

• Work closely with product, backend, data engineering, and platform teams to ensure successful integration and delivery.
  • Contribute to design discussions, technical documentation, and best practices for GenAI application engineering.