Headline

I get the job done.

Location

Pune District

Work experience

Computer Vision Research Engineer at Thelios AI, Jul 2023 - Present · 3 yrs 2 mos, Pune, Maharashtra, India

Computer Vision Engineer at Thelios AI, Jul 2023 - Present · 3 yrs 2 mos, • Led end-to-end development of sports video understanding pipelines, improving model accuracy from 70% to 98% through architecture optimization, temporal modeling, and data pipeline improvements. • Designed and implemented ball detection, player tracking, team-side detection, and rally extraction systems for large-scale, multi-camera sports video data. • Built action recognition models for basketball and volleyball using Video Swin Transformers and SlowFast Networks, capturing long-range temporal dependencies. • Addressed small-object detection challenges using multi-scale feature learning, high-resolution inference, and temporal frame stacking under motion blur and occlusion. • Developed ball-carrier identification and player–ball interaction modeling pipelines by integrating detection, tracking, and temporal context aggregation. • Conducted ablation studies and performance benchmarking across YOLOv8, SlowFast, and transformer-based spatio-temporal architectures. • Designed and implemented distributed, fault-tolerant ML pipelines using Prefect and MinIO for scalable data ingestion, inference, and evaluation. • Implemented pipeline state management for retries, intermediate artifacts, and data lineage tracking across distributed workflows. • Deployed production ML systems using Docker on AWS, managing the full lifecycle from research to deployment. • Mentored junior engineers and contributed to technical design reviews, improving ML engineering best practices and experiment reproducibility.

Education

Master's degree, Data science and ML, Indian Institute of Technology, Guwahati

Associate of Electrical Engineering, INSTITUTE OF TECHNOLOGY, GOPESHWAR

Public links

github: https://github.com/vimal-tomar

leetcode: https://leetcode.com/saitama_sensei