Manmeet Kumar

Generative AI Engineer focused on building production-grade LLM systems, agentic workflows, and RAG pipelines. Experienced in designing multi-agent architectures (AutoGen, LangGraph), fine-tuning models using PEFT (LoRA/QLoRA), and implementing evaluation-driven pipelines with guardrails and validation layers. Strong emphasis on execution, rapid prototyping, and deploying scalable AI systems in enterprise environments using Azure.

Data Scientist at Mercedes-Benz Research and Development India (2023-08 – Present)

Designed and deployed production-grade LLM systems automating engineering workflows, moving solutions from prototype to enterprise-scale deployment.

• Designed and deployed production-grade LLM systems automating engineering workflows, moving solutions from prototype to enterprise-scale deployment.
• Built multi-agent systems (planner–executor–validator) using AutoGen and LangGraph-style orchestration for solving complex, multi-step business logic.
• Developed hybrid RAG pipelines with embedding retrieval, chunking strategies, and re-ranking to improve factual accuracy and reduce hallucinations.
• Implemented guardrails, validation pipelines, and evaluation frameworks (automated + human-in-loop) to ensure reliable and safe LLM outputs.
• Optimized LLM pipelines for latency, token efficiency, and cost, enabling scalable and production-ready deployments.
• Co-inventor on a filed patent related to AI-driven automation of manufacturing planning workflows using LLM-based architectures.

B.Tech in Computer Science Engineering – VIT Bhopal (2019 – 2023)