I design and build real AI systems — agentic workflows, LLM pipelines, RAG architectures, and the ML infrastructure underneath. At HCLTech I shipped campaign systems to 20M+ users. On my own time, I build end-to-end: from model training to Kubernetes-deployed APIs with monitoring that pages me when something drifts.
Full architectures — not demos. Each project has a system design, real infrastructure, and production-grade monitoring.
An agentic system where specialised AI agents collaborate to research topics, synthesise findings, and produce structured reports. Built with LangGraph for orchestration and tool-use chains for web search, document analysis, and citation.
RAG pipeline that reads incoming tickets, classifies them, retrieves similar past resolutions via hybrid search, and drafts a response. Cut first-response time by 60% in testing.
Full production ML lifecycle: MLflow tracking, FastAPI inference, Docker + Kubernetes deployment, Prometheus/Grafana observability, and GitHub Actions CI/CD. The project that taught me how production ML actually works.
Production system at HCLTech. 40+ in-app campaigns during Black Friday/Cyber Monday reaching 20M+ users. Zero critical post-launch failures. Built the data pipelines powering personalised decisioning.
M.Tech AI/ML at BITS Pilani. Deep into agentic AI architectures, LLM fine-tuning, and multi-agent systems. Looking for roles where I can design and ship AI systems end-to-end — not just train models, but own the entire stack from architecture to deployment.