Computer Vision Engineer (YOLOv8 + Multi-Object Tracking) – Short Paid Sprint

I’m building a computer vision prototype that detects and tracks players in American football game film. This is a short, paid sprint (2–3 weeks) to ship working local inference — not a long-term role, not exploratory research. I already have: Labeled image frames (YOLO format) Clear scope and deliverables A locked data schema I’m looking for an engineer who can execute quickly and deliver a clean, working pipeline. Scope of Work (Phase 1) You will: Train / fine-tune YOLOv8 on provided labeled frames Implement multi-object tracking (ByteTrack, DeepSORT, or similar) Run inference on short video clips Output persistent player IDs per frame (visual overlay + CSV/JSON) This does not require: UI Cloud deployment Large-scale optimization Research papers Local inference is sufficient. Deliverables Working inference script (video → detections + tracked IDs) Example output video with bounding boxes + stable IDs Frame-level CSV/JSON with: frame_index object_id bounding_box (x, y, w, h) Brief README explaining how to run the pipeline locally Required Experience Proven experience with YOLOv5 / YOLOv8 Experience with multi-object tracking (ByteTrack, DeepSORT, Norfair, etc.) Comfortable working with video data Able to work independently and hit milestones Nice to Have (Not Required) Sports video or surveillance CV experience Experience cleaning or stabilizing tracker IDs Familiarity with Ultralytics ecosystem Timeline & Budget Timeline: 14–21 days Budget: Fixed price (milestone-based) Paid trial / Phase 1 only — no long-term obligation required How to Apply (Important) In your proposal, please include: A brief example of a YOLO + tracking project you’ve worked on Which tracker you’d recommend for this use case and why Your availability over the next 3 weeks Proposals that don’t address these will be ignored. Apply tot his job