At VEHANT TECHNOLOGIES PVT. LTD., I streamlined secondary detection architecture for real-time video surveillance, optimizing CPU and GPU resources across edge devices and servers. I’m also exploring model porting to NPUs and moving products from x86 to ARM.
At META PLATFORMS INC., deputed by CREOSPAN INC., I built distributed pipelines to analyze AR/VR telemetry, improving calibrated projection accuracy by 25%. I also fixed race conditions in multi-camera systems and deployed a robot arm for data capture.
My work has spanned computer vision research and production systems, from accelerating object detection training 40× at PENNSYLVANIA STATE UNIVERSITY to developing optimized video analytics, attendance, and ADAS algorithms. I’ve worked with tools including C++, Python, PyTorch, OpenCV, TensorRT, and Docker.

