Computer Vision Engineer
Core
Build and improve sports video intelligence models covering detection, tracking, pose estimation, event understanding, and multi-view reasoning.
Role type
Applied Computer Vision Engineer (Research + Production)
Builds
Sports video intelligence models and scalable inference pipelines
Domain
Sports analytics, Computer Vision, Video Processing
Deliverable
production ML models
Required skills
PyTorch, Python, video CV fundamentals (occlusion, temporal consistency, identity tracking), model training and debugging, data pipeline engineering, experiment design and evaluation
Preferred skills
MLOps, model serving (Triton/TorchServe), FFmpeg, WebDataset, transformer architectures, weak/self-supervision techniques
Technologies
PyTorch, Python, FFmpeg, WebDataset, Triton, TorchServe
Responsibilities
Build and train CV models for player/ball detection, multi-object tracking, and pose estimation; Own the experimentation loop including ablations and error analysis; Design evaluation metrics and datasets; Improve data efficiency via augmentations and sampling strategies; Prototype modern architectures; Collaborate on dataset labeling design; Help productionize models for batch/stream inference; Add quality gates for reproducibility and regression detection
Seniority
Mid to Senior, scope-dependent
