Built scalable, production-grade GenAI solution automating document extraction and comparison for banking and finance using Azure OCR, custom chunking, and LLM prompting, reducing manual effort by 30-40%

Developed auto prompt optimization feature using DSPy to automate prompt generation from input-output pairs, eliminating user effort on complex prompt writing

Built KYC/AML agent automating investor profiling by orchestrating tools through MCP server, improving onboarding efficiency and risk assessment accuracy

Led design, development, deployment and maintenance of Video Motion Detection product using YOLO and Python to detect real-time people intrusion, with full MLOps integration using MLflow and Kubeflow

Led design and development of RAG-based QA chatbot using OLLAMA, LangChain, and Python, replacing traditional FAQ chatbot and enhancing user experience

Developed and deployed containerized AI microservices for real-time video intelligence including camera tampering and facial spoofing detection on Kubernetes

Worked on end-to-end Computer Vision projects including object detection, image segmentation, and classification models, deployed on AWS and edge devices

Developed AI use-cases including Vending Machine Recommendation Engine, Mispicks Reduction in warehouses, and Safety Alert Automation using Computer Vision and Python

Optimized training and inference pipelines through distributed training, false positive reduction, and Nvidia packages including Tao/TLT and Deepstream for process and inference optimization

Built Course Recommender System suggesting courses from Coursera and Udemy based on user preferences

Predicted similar skills using trained Word2Vec model on LinkedIn skills dataset to match user profile skills

Implemented recommendation system using Gensim library for word embedding and similarity analysis