Senior Computer Vision Engineer - Object Detection, Tracking & Sports Video Analytics

Job Title: Senior Computer Vision Engineer — Object Detection & Sports Video Analytics Location: U.S.-based candidates only / Remote Type: Early-stage startup role Compensation: Equity + Base Salary. To be discussed based on experience, commitment level, and fit About Virtus Virtus is building an AI-powered sports visibility and sponsorship intelligence platform that helps athletes, universities, teams, brands, and agencies measure the real commercial value of athlete exposure. Our platform uses computer vision and video intelligence to detect, track, and measure athlete visibility across broadcast, streaming, social, and digital video. We turn raw footage into objective proof of exposure, including screen time, visibility quality, brand presence, and sponsorship value. Role Overview

• Virtus is looking for a hands-on Senior Computer Vision Engineer specializing in object detection, tracking, and sports video analytics.
• This is not a general AI role. We are specifically looking for someone with direct experience building, training, fine-tuning, evaluating, and deploying computer vision models for object detection, multi-object tracking, recognition, and video analysis.
• The right candidate should have hands-on experience with models and frameworks such as YOLO, Faster R-CNN, Detectron, OpenCV, ByteTrack, DeepSORT, BoT-SORT, OCR, segmentation models, or similar tools.
• For Virtus, this means helping us detect and track athletes, jersey numbers, logos, brand marks, sponsor exposure, and athlete screen presence across real-world sports video.

What You Will Work On

• You will help build the computer vision systems that allow Virtus to:
• Detect and track athletes in sports video.
• Measure athlete screen time and visibility quality.
• Detect jersey numbers, uniforms, logos, sponsor marks, and brand exposure.
• Convert raw video into structured exposure data.
• Generate visibility metrics, confidence scores, and analytics-ready outputs.
• Support athlete media kits, NIL reporting, sponsorship valuation, and partner API outputs.

Key Responsibilities

• Build, train, fine-tune, evaluate, and deploy computer vision models for object detection and video analytics.
• Develop pipelines for athlete detection, player tracking, jersey/number recognition, logo detection, brand exposure detection, and screen-time measurement.
• Work with sports video from broadcasts, social media clips, highlight reels, and digital content.
• Improve model performance across real-world conditions such as motion blur, fast camera movement, camera cuts, partial occlusion, replays, low-resolution clips, crowded scenes, and changing lighting.
• Build and optimize inference pipelines for speed, cost, and accuracy.
• Write clean, production-ready Python code for training, testing, inference, evaluation, and deployment.
• Prepare and manage datasets, annotations, bounding boxes, class definitions, and model validation workflows.
• Evaluate model performance using metrics such as precision, recall, mAP, IoU, confidence thresholds, false positives, false negatives, and tracking accuracy.
• Collaborate with product, backend, and leadership teams to connect model outputs to dashboards, reports, media kits, and APIs.

Required Experience

• Deep hands-on experience in computer vision, especially object detection and video analysis.
• Strong Python programming skills.
• Experience with object detection models such as YOLO, Faster R-CNN, SSD, RetinaNet, Detectron, MMDetection, or similar.
• Experience with YOLO-based workflows such as YOLOv5, YOLOv7, YOLOv8, YOLOv9, YOLOv10, YOLOv11, Ultralytics, or similar.
• Experience with video-based computer vision, not only static image classification.
• Experience with multi-object tracking approaches such as DeepSORT, ByteTrack, BoT-SORT, OC-SORT, or similar.
• Experience with OpenCV, PyTorch, TensorFlow, or similar frameworks.
• Experience preparing datasets for computer vision models, including annotation workflows, labeling, bounding boxes, class definitions, and quality control.
• Experience optimizing inference performance in GPU-based environments.
• Ability to work with noisy, imperfect, real-world video data.
• Comfortable working independently in a fast-moving startup environment.

Preferred Experience

• Sports video analytics.
• Athlete/player detection and tracking.
• Logo detection, brand exposure detection, sponsorship measurement, or media analytics.
• OCR or visual text recognition for jersey numbers, signage, boards, scoreboards, or broadcast graphics.
• Segmentation models such as M

Apply To This Job