Senior Computer Vision Engineer at OZ Sports (2023-07 – Present)
Multi-Camera AI Soccer Production System and AI-Powered Fish Cutting Automation System
- Designed and implemented an end-to-end multi-camera computer vision pipeline for automated sports production and analytics.
- Developed a six-camera vision system for detecting and tracking players, referees, and the ball in live soccer matches.
- Implemented real-time object detection and tracking pipelines using YOLOv11, ByteTrack, and DeepSORT, achieving stable tracking performance at 60 FPS.
- Designed multi-camera calibration pipelines including intrinsic and extrinsic parameter estimation using geometric ray projection and multi-view consistency optimization.
- Implemented multi-view 3D ball localization and trajectory reconstruction using triangulation and temporal filtering.
- Developed pitch landmark detection and spatial reasoning modules for contextual understanding of the game environment.
- Optimized the perception pipeline for real-time deployment on NVIDIA Jetson AGX Orin.
- Conducted system validation and performance tuning for real-world broadcast scenarios.
- Enabled automated sports analytics including ball tracking, player tracking, and event detection.
- Built a production-ready perception pipeline capable of operating in dynamic outdoor sports environments.
- Explored LLM- and NLP-driven orchestration approaches for camera control and higher-level automated production decision systems.
- Developed a deployable AI + robotics perception system for automated fish processing.
- Served as project lead, owning the end-to-end lifecycle from client requirement gathering through system development, integration, and final product delivery.
- Built a vision pipeline combining object detection, segmentation, and depth sensing using Intel RealSense and ZED cameras.
- Implemented volume and weight estimation from depth maps for precision robotic cutting.
- Designed a keypoint-based prediction model for optimal cut positioning.
- Integrated perception outputs with a Fanuc industrial robot using optimized motion planning algorithms.
- Deployed the full AI pipeline on Jetson AGX Orin for low-latency edge inference.
- Led and guided the data collection and annotation workflow, coordinating the team to build high-quality datasets for model development and validation.
Research Scholar (Ph.D.) at VNIT Nagpur (2019-01 – 2023-01)
Department of Electronics & Communication Engineering - Research Focus: Computer Vision and Deep Learning
- Developed novel architectures for salient object detection and segmentation using deep learning and spiking neural networks.
- Designed boundary-preserving detection frameworks improving segmentation accuracy in complex scenes.
- Published multiple papers in IEEE, Springer, and Elsevier journals.
Assistant Professor at RCOEM Nagpur (2016-01 – 2019-01)
Electronics & Communication Engineering Department
- Taught courses in Machine Learning, Digital Signal Processing, and Embedded Systems.
- Guided undergraduate research projects in computer vision and artificial intelligence.
