Computer Vision Engineer
About the project
The company films youth football matches with its own cameras and wants to fully automate match analysis, scaling to about 100,000 games per month, a volume that manual tagging can’t support. The system needs to automatically detect events (passes, shots and other actions), link them to the right player, pitch location and even body part (for example, which foot touched the ball), and track every player’s position throughout the whole match.
The main challenge is accuracy: existing solutions reach only 60–80%, while the target is 98–99%. The hardest part is correctly identifying players after substitutions, especially when they happen off camera or at halftime, as well as football-specific situations like disallowed goals or play being brought back by the referee.
Requirements
- Strong experience in multi-object tracking and player re-identification in sports video (re-identification is the key to solving substitutions).
- Pose / keypoint estimation for body part detection.
- Action and event recognition from video.
- Proven experience improving models to production-grade accuracy at high volume.
- Experience with football or other team sports is a strong plus.
Tech stack
Must-have:
- Python, PyTorch
- OpenCV and video processing (FFmpeg)
- Multi-object tracking and re-identification models
- Pose / keypoint estimation models
- Video-based action and event recognition models
- Camera calibration and homography (mapping video to pitch coordinates)
- GPU training and inference, model optimization for production (e.g., TensorRT, ONNX)
Nice to have:
- Tracking: ByteTrack, DeepSORT, BoT-SORT
- Pose estimation: HRNet, ViTPose, MMPose
- Action recognition: SlowFast, VideoMAE
- Experience with SoccerNet datasets or challenges
- Cloud and MLOps for large-scale video processing (AWS / GCP, Docker, Kubernetes)
What we offer:
- Covered vacation period: 20 business days and 8 days off
- Free English classes
- Flexible working schedule
- Truly friendly and supporting atmosphere
- Working remotely or in one of our offices
- Medical insurance for employees from Ukraine
- Legal support