Shivansh Tripathi

I've built AI-powered systems spanning real-time computer vision, fraud detection, and GenAI evaluation. My work combines Python, machine learning, FastAPI backends, and scalable AWS deployments.

For an AI-powered smart surveillance system, I used YOLOv8 and OpenCV for human detection and activity monitoring, then designed an event-driven FastAPI backend for real-time alerts. I deployed the pipeline on AWS using EC2, S3, Lambda, and SNS.

Through freelance work with Micro1 and Outlier, I've contributed to AI training workflows, frontend tasks, data quality, and model outputs. I've also built fraud-detection and AI evaluation platforms using PR-AUC optimization, SHAP, Docker, and rule-based and LLM-as-a-judge evaluation approaches.