I design and deliver production multi-agent systems โ from LangGraph orchestration and MCP server development, to system prompt engineering and agent evaluation pipelines. I care about agents that work in the real world, not just in demos.
Deep specialization across the full lifecycle of production AI systems
Designing and shipping production multi-agent pipelines using LangChain and LangGraph โ with robust state machines, tool routing, and failure handling.
Building custom MCP (Model Context Protocol) servers and integrating enterprise tools โ giving LLMs structured, reliable access to real systems.
Crafting production-grade system prompts โ with precise output formatting, structured reasoning chains, and versioned prompt management.
Designing eval frameworks that catch regressions before production โ LLM-as-judge pipelines, tracing, scoring, and behavioral reliability testing.
End-to-end RAG pipelines โ from embedding strategy and chunking to retrieval optimization and vector store management at scale.
Production-grade FastAPI backends and cloud deployments on AWS โ containerized, CI/CD-ready, and built to scale with AI workloads.
Production AI systems delivering measurable impact
Automated invoice processing using multimodal GenAI with a conversational interface โ cutting processing time from hours to minutes per batch.
24/7 AI-powered customer assistant for businesses โ trained on real website content, with real-time knowledge base updates and natural conversation flow.
Enterprise ML pipeline for supply chain optimization โ 92% accuracy, containerized deployment with full CI/CD.
RAG chatbot integrated with Slack โ instant access to org knowledge with 95% retrieval accuracy and 1000+ queries/week.
AWS Bedrock conversational AI for sales automation โ full serverless architecture built and deployed in 48 hours.
Tools and technologies I use to build production AI systems
Senior AI Engineer with 3+ years building and deploying intelligent systems. My core focus is production multi-agent architecture โ specifically agent orchestration, MCP server development, precise prompt engineering, and the eval frameworks that make it all trustworthy enough to ship.
I've shipped everything from invoice-parsing chatbots and enterprise RAG systems to Slack-integrated knowledge assistants and cloud-native AI agents on AWS Bedrock. I care deeply about the gap between "it works in a notebook" and "it works in prod" โ and I've spent years closing that gap.
Every system I design accounts for failure modes, deterministic routing, and observability from day one.
Built custom MCP servers integrating Jira, Confluence, GitHub, Grafana, and LMS platforms for real enterprise workflows.
I don't ship agents without evals. LLM-as-judge, regression suites, and tracing are part of every delivery.
Open to remote Senior AI Engineer roles and interesting agent / LLM infrastructure projects
(+91) 9164623536