I architect agentic AI systems that ship to production. _
7+ years in software engineering, 2+ of them building and running large language model systems — multi-agent orchestration, retrieval-augmented generation, and LLM evaluation at scale.
Building reliable AI systems, not demos
I'm a GenAI & Agentic AI Engineer with 7+ years of software engineering experience, including 2+ years designing and shipping production large language model systems. My focus is the part that's hard: making agents reliable, observable, and cost-efficient once they leave the notebook.
At NTT Data I architected a LangGraph multi-agent platform powering enterprise applications used by thousands of users, built hybrid RAG pipelines over 10K+ documents, and stood up an LLM evaluation and safety framework that catches regressions before they reach users. Before AI, I spent years as a full-stack engineer at Oracle and beyond — which is why my systems come with real APIs, tests, and CI/CD.
I'm currently pursuing an M.Tech in Artificial Intelligence at IIT Jodhpur, and I'm open to roles in Agentic AI, Generative AI, LLM, and AI/ML engineering.
Technical stack
agentic ai
llm frameworks & platforms
rag & vector stores
evaluation & observability
languages & backend
data & cloud / devops
Where I've worked
- Architected a production multi-agent platform on LangGraph with ReAct reasoning, intent routing, persistent DynamoDB memory and Redis caching — cutting median handling time 26% at 96%+ response quality.
- Built a hybrid RAG pipeline over 10K+ documents (BM25 + dense k-NN on Amazon OpenSearch, Cohere re-ranking) with structure-aware chunking and metadata filters, reaching 90%+ recall@5 on an SME golden set.
- Architected a regulated-grade insurance claim-adjudication pipeline on AWS — LangGraph orchestration, RAG (Amazon Textract + Bedrock) for policy-rule extraction, and a deterministic rules engine writing auditable verdicts to an append-only Aurora PostgreSQL trail.
Python · LangGraph · LangChain · MCP · A2A · AWS Bedrock · Amazon OpenSearch · Textract · Aurora · FastAPI · DynamoDB · Redis · Docker · Kubernetes
- Integrated REST APIs and real-time clinical data pipelines into Cerner's EHR platform, enabling live patient-data retrieval across ML-assisted diagnostic workflows with 67% adoption across clinical teams.
- Led platform-wide browser modernization (IE → Edge Chromium), cutting rendering time 85% and CPU usage 33%.
- Built CI-integrated testing with React Testing Library, enabling teams to reach 100% critical-path coverage.
- Awarded the Rising Star Award for delivering the Edge Chromium migration two months ahead of schedule.
React · TypeScript · JavaScript · REST APIs · Node.js · CI/CD
- Engineered real-time chat and search with persistent history for high-traffic applications, lifting user engagement 10%.
- Rebuilt the core transaction flow as a React + Node.js SPA (replacing a legacy multi-page app), cutting processing time 40% and improving retention 25%.
- Created a Node.js production diagnostics module that improved stack-trace quality and cut debugging time for the team.
React · Node.js · JavaScript · REST APIs
Selected work
Multi-Agent AI Platform
Production Agentic AI platform with ReAct reasoning, intent routing, persistent memory, tool calling and quality-gated sub-task decomposition. Reduced median handling time 26% at 96%+ quality.
Hybrid RAG Knowledge System
Retrieval-augmented generation over 10K+ documents with BM25 + dense k-NN on Amazon OpenSearch and Cohere re-ranking — 90%+ recall@5 on an SME golden set.
Regulated Claim Adjudication
Auditable insurance claim-adjudication pipeline on AWS: LLM extraction (Amazon Textract + RAG) with a deterministic rules engine, every verdict written to an append-only Aurora PostgreSQL audit trail.
Credentials
M.Tech — Artificial Intelligence
B.Tech — Computer Science & Engineering
Certifications
- Microsoft Certified: Azure AI Engineer Associate
- Oracle Cloud Infrastructure — Certified Generative AI Professional
- Oracle Cloud Infrastructure 2023 — Certified Foundations Associate
Let's build something intelligent.
Open to Agentic AI, Generative AI, LLM and AI/ML engineering roles. The fastest way to reach me is email or LinkedIn.