At Genpact, I built an enterprise multi-agent AI platform that routes requests across specialized agents, including RAG agents. I implemented LangGraph orchestration, MCP and API-based tools, and retrieval workflows connected to AWS Bedrock and enterprise systems.
I also developed an insurance RAG solution that maps natural-language underwriting requirements to Property IDs and Object IDs. I built its ingestion pipeline to extract information from HTML, generate embeddings, and index knowledge in MongoDB Atlas Vector Search.
For enterprise applications running on AWS, I implemented AI/LLM observability and application performance monitoring. I integrated Datadog and Arize with OpenTelemetry and ddtrace to capture traces, LLM interactions, latency, errors, and performance metrics.
