Vishal User

I’m an AI/ML Engineer with 1+ year of hands-on production experience building enterprise LLM agents, multi-agent orchestration systems, RAG pipelines, and scalable backend microservices. I delivered 2 end-to-end AI products for a Fortune 500 energy client at LTIMindtree.

In my recent role, I designed and deployed 4 production LLM agents on Azure AI Foundry (Incident, Problem, Change, Service Request) integrated with ServiceNow and Jira. I also built secure REST APIs, webhooks, and async event-driven integrations—always with a strong focus on reliability and safe operations.

I engineered end-to-end RAG systems, including document ingestion, chunking strategies, hybrid vector search, and response generation using Azure AI Search. I extended beyond standard retrieval with Knowledge Graph RAG, Adaptive RAG, Multimodal RAG, and cross-encoder re-ranking, then validated quality using LangSmith and RAGAs to improve context recall and answer faithfulness.

I’m particularly proud of performance and cost wins: I reduced 50,000+ individual LLM API calls to ~500 async batches using Azure Storage Queues, cutting inference cost by ~90%. I also resolved a Microsoft Defender false-positive caused by a misconfigured RAG retry loop (zero downtime), and I bring that same production-debugging mindset to every agentic workflow. I’m also familiar with Hugging Face models, prompt engineering, open-source inferencing (vLLM), and MLOps workflows.