Headline

“Senior GenAI & Agentic AI Engineer | LangGraph · Agentic AI · MCP · RAG Evaluations · LLM Evaluations · React ·Typescript | Built Systems Handling 800K+ Daily Requests @ Optum”

About

I am a GenAI & Agentic AI Engineer with 3+ years of experience building, evaluating, and deploying production-grade AI systems at enterprise scale. At Optum (UnitedHealth Group), I specialize in designing multi-agent LangGraph workflows, high-throughput FastAPI backends, and advanced RAG architectures that process over 800,000+ daily production requests. My core focus is bridging the gap between cutting-edge LLM capabilities and reliable, production-ready backend systems. Core Technical Expertise: • Agentic AI & Orchestration: Multi-Agent LangGraph Workflows, LangChain, Tool & Function Calling, Prompt Engineering • RAG & Search Systems: Dense Embeddings, Semantic Chunking, Hybrid Search, BERT Cross-Encoder Reranking, ChromaDB, PostgreSQL (pgvector) • Custom LLMOps & Evaluation: Programmatic Benchmarking Frameworks, Precision/Recall/F1 Tool Validation, Model Observability, LangFuse and OpenTelemetry • Backend & Microservices: Python, FastAPI, Node.js, Distributed Systems, Event-Driven Architecture, Async REST APIs • Cloud & Security: Azure OpenAI, AWS (EC2, S3, Lambda), Docker, OWASP Security, OAuth 2.0 / JWT Currently based in Gurugram, India and open to high-impact roles locally, as well as overseas relocation opportunities in Singapore and UAE.

Location

Gurugram

AI stack

langchain

langfuse

langgraph

mcp

Pinecone

PydanticAI

Work experience

Senior Software Engineer - GenAI and AI Systems at Optum, Aug 2026 - Present · 3 mos, Gurugram, Haryana, India · Hybrid

Software Engineer - GenAI & AI Systems at Optum, May 2023 - Aug 2026 · 3 yrs 4 mos, Gurugram, Haryana, India · Hybrid, • High-Scale Multi-Agent Orchestration (MCP): Engineered and deployed production multi-agent workflows using LangGraph / LangChain integrated with MCP standards, Azure OpenAI, and FastAPI backends; applied prompt versioning and dynamic model routing across an enterprise platform processing 800,000+ daily requests. • Commerce Workflow Automation Safety Guardrails: Automated complex insurance sales and commerce transaction workflows by engineering autonomous AI agents with custom safety guardrails and Human-in-the-Loop (HITL) approval gates, eliminating multi-screen navigation and reducing operational processing effort by 70%. • Enterprise RAG Architecture Grounding: Built an advanced RAG pipeline for unstructured/structured insurance policy data using semantic chunking, dense embeddings, hybrid retrieval, and BERT Cross-Encoder reranking on pgvector / ChromaDB , decreasing decision-making time by 35%. • Custom LLMOps Enterprise Observability Transition: Architected a code-based custom evaluation engine using golden datasets to validate tool selection accuracy and extraction Precision, Recall, and F1-Scores; scaled production reliability by transitioning from custom evaluation harnesses into an enterprise Langfuse and OpenTelemetry (OTel) tracing and observability stack. • Multimodal Document Verification (KYC/Onboarding): Owned end-to-end development of a real-time document validation microservice using multimodal OCR to verify uploaded customer records and onboarding documents during live sales transactions, reducing manual effort by 60%. • Real-Time Agent Interfaces Streaming: Built responsive streaming UI components in React / TypeScript integrated asynchronously with backend LangGraph engines to deliver low-latency, multi-turn agentic interaction.

Education

Bachelor of Technology - BTech, Computer Science, Kalinga Institute of Industrial Technology, Bhubaneswar

Public links

github: https://github.com/Dmishra13