At Tulapi.ai, I contributed to applied machine learning and computer vision workflows using Python, TensorFlow, OpenCV, and FastAPI. I also built model-training, evaluation, and custom-image testing workflows.
I built EmotionLens, a real-time app that tracks two people on live video and flags possible hits. Its learned model raised punch-clip detection from 4/10 to 8/10 and cut false alerts from 6/20 to 2/20 on public LIMU clips.
I also built VideoGit, a local-first desktop app that turns DaVinci Resolve and Kdenlive timelines into clip-level changes; it won 1st Place at DevJams ’26. As Co-Secretary of ACM-VIT, I deploy and maintain production systems and mentor junior members.

