Purnima

Gen AI/ML Engineer

Larsen & Toubro • Período integral

Apr 2025 - Present • 1 yr 2 mos

I work as a Generative AI & ML Engineer at a large-scale engineering enterprise, where I design and deploy production-grade AI systems used by enterprise teams daily. My work spans the full AI lifecycle — from architecture and model integration to cloud deployment and real-time monitoring. I build and deploy LLM-powered chatbots on Azure Kubernetes Service (AKS) with CI/CD pipelines, achieving 99% uptime and reducing deployment time by 40%. I engineer end-to-end RAG pipelines using LangChain and Groq, integrating Pinecone and FAISS vector databases for semantic search across large enterprise document sets — designing chunking strategies, embedding pipelines and retrieval optimisation to ensure accurate, contextually relevant responses at scale. I built an Agentic AI workflow integrated with Microsoft Teams that autonomously converts meeting transcripts into structured tasks with deadlines and ownership tracking — awarded Gold Medal at a National-level Technical Workshop. I also engineered an Automatic Schema Mapper using FastAPI and LLM-based fuzzy matching that improved data processing speed by 40% and reduced manual cleaning effort by 60%. As an ML intern at the same organisation, I built a schema-based LLM chatbot enabling non-technical users to query structured databases in plain English — achieving 15% higher contextual accuracy over standard RAG approaches. Contributed to ETL preprocessing, instruction design, data loading logic and full application architecture. I also design monitoring and logging frameworks that track model drift, latency and errors in real time — ensuring AI systems stay reliable long after launch. Every solution I deliver is clean, documented, scalable and built for real users — not just demos.