Headline

AI Engineer | Generative AI · LLMs · Agentic AI · RAG · AWS Bedrock · LangChain | Production LLM Platforms for Enterprise

About

I build AI systems that work in production — not just demos. Over the past 4 years I've shipped Generative AI applications, RAG pipelines, agentic workflows, and LLM-powered automation across insurance document intelligence, automotive CRM personalization, and enterprise knowledge management. My work spans the full stack — from raw document ingestion and vector search to structured, citation-backed, audit-ready outputs that real compliance analysts and field technicians depend on daily. What I've shipped: → Insurance document intelligence platform (AWS Bedrock + AgentCore + LangGraph) — cut manual compliance review effort by ~60%, analyst turnaround from 2–3 days to minutes → Automotive CRM personalization engine (AutoGen + LangChain + RAG) — improved engagement response rates by ~35%, deployed in production serving thousands of leads → Machine manual AI assistant (OCR + FAISS + Semantic Search + FastAPI) — won 2nd place in a 500+ company hackathon, production-deployed for factory technicians → HR policy assistant (LLaMA + LangChain + On-Premise RAG) — reduced HR query resolution effort by ~40% within the first month of deployment My stack: ChatGPT/OpenAI · Claude · Gemini · AWS Bedrock · AgentCore · LangChain · LangGraph · AutoGen · CrewAI · LLaMA · FAISS · ChromaDB · Pinecone · Weaviate · FastAPI · Docker · Kubernetes · Python · PyTorch · RAGAS · LangSmith · REST APIs · Microservices · CI/CD I specialize in the parts of GenAI that most engineers avoid: hallucination control, citation-backed outputs, strict prompt contracts, semantic search quality, AI governance, and audit-ready pipelines for regulated industries. If it needs to work reliably at enterprise scale — that's where I focus. Open to AI Engineer, Senior GenAI Engineer, or Agentic AI Engineer roles (Remote / Mumbai). Connect or DM.

Location

Mumbai

AI stack

antigravity

cursor

claudecode

Work experience

AI ML Engineer at OverseeAI, Jul 2025 - Present · 1 yr 2 mos, Mumbai, Led end-to-end design and deployment of a production-grade GenAI insurance document intelligence platform using ChatGPT/OpenAI, Claude, and AWS Bedrock — comparing internal, state, competitor, AAIS, and ISO forms to generate audit-ready compliance outputs including Change Logs, Regulatory Impact summaries, and underwriting decision support. - Architected autonomous multi-agent workflows using AWS AgentCore and LangGraph to fully automate insurance forms comparison, ratebook generation, gap detection, and structured compliance output generation — cutting analyst turnaround from 2–3 days to minutes and reducing manual effort by ~60%. - Engineered scalable production RAG pipelines with custom chunking, metadata extraction, document-level filtering, and semantic search over enterprise data sources — improving answer grounding, reducing hallucinations, and maintaining auditability for regulated compliance workflows. - Designed citation-backed LLM output flows where every generated finding is traceable to source filing text, PDF pages, snippets, and document labels — meeting AI governance, security, and data privacy requirements for insurance operations teams. - Built reusable prompt engineering frameworks with strict output contracts for structured tables, JSON blocks, change logs, section-wise comparisons, and deterministic response formatting — achieving consistent structured output across 95%+ of generations. - Integrated AI models with enterprise applications via REST APIs and microservices; collaborated with stakeholders to define AI use cases, product workflows, and reviewer-friendly output experiences for compliance and underwriting teams.

Machine Learning Engineer at WonderBiz Technologies, Jun 2022 - Feb 2025 · 2 yrs 9 mos, Mumbai, - Won 2nd place in a 500+ company hackathon — built a GenAI enterprise AI assistant for factory technicians using OCR, FAISS-backed semantic search, and LLM Q&A over multi-format machine manuals and condition-based formulas. Deployed via FastAPI microservice with sub-second retrieval latency under concurrent load. - Built a GenAI CRM personalization engine using OpenAI LLMs, AutoGen, LangChain, and RAG — generating hyper-personalized customer emails and WhatsApp messages conditioned on lifecycle stage, persona, purchase history, transcript context, and business rules. Deployed live in an enterprise automotive CRM platform serving thousands of leads, improving engagement response rates by ~35%. - Implemented a multi-agent generator–evaluator loop (AutoGen) where one LLM agent generated personalized messages and a critic agent scored quality, tone, relevance, and grounding — reducing manual QA cycles by over 50%. - Built a self-hosted RAG-based HR policy assistant using LLaMA and LangChain over internal company documents — on-premise ingestion, chunking, FAISS vector indexing, and full RAG pipeline meeting strict data privacy and AI governance requirements — reducing HR query resolution effort by ~40% within the first month. - Engineered context-aware, reusable prompt templates with dynamic injection of customer attributes, behavioral history, product interest, and RAG-retrieved enterprise context — enabling personalized LLM output at scale without fine-tuning.

Education

Electronics and Telecommunication, Telecommunication, Vidyalankar Institute of Technology, Mumbai

Diploma, Electronic and Telecommunications, St. Xavier's Technical Institute