At Accenture, I designed and built the conversational layer of a GenAI chat application on AWS using Amazon Bedrock. I implemented RAG and SQL RAG so people can ask natural-language questions and receive grounded answers from documents and structured data.
At Audax Labs, I designed an OCR and data extraction system for financial documents using Qwen2.5-VL and classical machine learning. I also developed a smart bus safety system on NVIDIA Jetson Nano to detect standing students and track boarding and deboarding counts.
At WalkingTree Technology, I built a stock analysis agent with LangGraph and automated strategy-based code generation and execution with AutoGen. Earlier, at Audax Labs, I developed a RAG chatbot for an electric bike manufacturer that reduced customer care costs by 20%, and a patient-assessment chatbot that reduced intake time by 30%.
My projects include a multimedia RAG assistant using Llama 3.1, Whisper, and ChromaDB, evaluated at 98% faithfulness and 92% answer correctness with Ragas. I’ve also built deep-learning systems for image classification and real-time object detection.
