Sarvesh Harikant

I’m an AI Developer focused on building retrieval-augmented and document-centric LLM systems that produce precise, domain-ready answers. I specialize in RAG applications, generative extraction, and document preprocessing to improve content retrieval quality.

At Eximietas Design, I built a RAG-powered chatbot for specialized question answering (including VLSI and PCIe) and implemented Hybrid RAG and Naïve RAG architectures to enhance retrieval precision and response generation. I also developed multimodal AI models that integrate image and text processing, structuring extracted content for efficient storage and retrieval.

I design agentic LLM frameworks leveraging REACT prompting to improve reasoning, decision-making, and workflow automation, and I fine-tune open-source LLMs with LoRA for production-ready tasks like resume and profile generation. Previously, as a Computer Vision Engineer, I delivered real-time object detection and segmentation systems and improved accuracy (20%), reduced false positives (30%), and accelerated training/deployment efficiency (40% decrease in training time), while optimizing segmentation complexity by 15% without sacrificing precision.