Open to opportunities

Prem Sagar User

@premsagariitb

Message

I build production LLM, RAG, and agentic AI systems for insurance, consulting, and real estate.

India

Message

What I'm looking for

I'm open to GenAI and LLM Engineer roles in Bangalore, Hyderabad, Mumbai, or Pune, where I can build production RAG, agentic AI, and Vision-Language Model systems.

At Niva Bupa, I own the Document AI track for Claims 2.0, replacing legacy OCR with self-hosted Vision-Language Models for 100K+ health-insurance claims each month. I also design an agent-based underwriting copilot with human-in-the-loop sign-off and automated decisioning.

Previously, at KPMG, I architected a multi-agent RAG pipeline using GPT-4o, Semantic Kernel, and Azure AI Search across 10,000+ pages, achieving 95% relevance and reducing analysis time by 90%. At RealtyAI, I built a Gemini-powered resume-screening engine that cut screening time by 70% and deployed it to AWS EC2 through CI/CD.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Niva Bupa Health Insurance

Jun 2026 - Present (4 months)

• Architecting a three-agent Claims 2.0 platform (Document AI, Financial AI, Medical AI), owning the Document AI track end-to-end for a workflow processing 100K+ health-insurance claims/month
• Replacing legacy OCR (AWS Textract) with self-hosted Vision-Language Models for scanned medical-claim extraction, converting unstructured documents into standardized structured JSON
• Designing an agent-bas

LangGraph Vector Databases Ragas Model Context Protocol (MCP)

Deal Analytics-AI

KPMG

May 2025 - Aug 2025 (3 months)

• Orchestrated multi-agent RAG pipeline with Semantic Kernel, leveraging GPT-4o (vision & LLM), architecting multi-modal RAG with chunking & CoT/Few-Shot prompting to slash project timelines by over 70% for insights
• Engineered high-precision retriever over 10k+ pages, using Azure AI Search hybrid search for 95% relevance
• Deployed an AI Platform on Azure with FastAPI & PostgreSQL for metadata t

Vector Databases Semantic Kernel fastAPI Generative AI SQL Agent Machine Learning AI PostgreSQL GPT 4 Deal Strategy

GenAI Intern

RealtyAI

Dec 2024 - Apr 2025 (4 months)

• Engineered high-fidelity character consistency using Flux face-matching in ComfyUI, enforcing 90% cosine similarity
• Orchestrated a generative pipeline using GPT-4o for 20+ scene narratives & TensorART for scene image synthesis
• Architected AI resume screening engine using LovableAI, integrating Gemini AI, slashing screening time by 70%
• Deployed the containerized AI application system to AWS

CI CD Pipelines Vector Databases Machine Learning Docker ComfyUI Image Generation MongoDB