Computer Vision Engineer

Department: EngineeringReports To: Computer Vision LeadLocation: Austin, TX or Los Angeles, CAAbout the RoleWINs computer vision team turns video into sports insights. Our models extract data from game and training footage captured by thousands of Smart Park camera systems at baseball and softball fields and turn it into stats, metrics, and insights that help players, coaches, and teams get better.As a Computer Vision Engineer, youll build and ship the models and pipelines behind those insights: detecting and tracking the ball and players, measuring what happened on every pitch and swing, and doing it reliably across real-world fields, lighting, and camera angles.This is a role for an engineer who has shipped computer vision in commercial products used by real customers not just research or demos. Success is measured by the accuracy, reliability, and value of the data we deliver.Key ResponsibilitiesModels & PipelinesDesign, train, and deploy models for object detection, multi-object tracking, pose estimation, and event recognition from sports video.Build end-to-end pipelines that turn raw video into structured, reliable sports data ball flight, player movement, and swing and pitch mechanics.Optimize models for production: accuracy, latency, cost, and edge vs. cloud inference.Data & EvaluationBuild datasets, labeling workflows, and evaluation frameworks that measure accuracy against ground truth.Monitor model performance in the field across thousands of cameras, varied lighting, field conditions, and camera placements.Find failure modes fast and close them with better data, better models, or better calibration.Shipping ProductWork with product, backend, and mobile engineers to turn model outputs into features customers use every day.Partner with hardware on camera selection, placement, calibration, and synchronization.Stay current on research and bring the right advances into production.Team & StructureIndividual contributor on WINs computer vision team, reporting to the Computer Vision Lead.Close partners: data and backend engineering, product, hardware, and Live Ops.Who You AreYouve shipped. Your models run in commercial products used by real customers, and you know what it takes to get them there and keep them working.You go deep. You have strong fundamentals in deep learning, tracking, and multi-view geometry, and you can reason from first principles when off-the-shelf approaches fall short.You care about accuracy you can prove. You measure against ground truth, and you dont trust a metric you cant validate.You love sports. You know what spin rate and exit velocity mean, and youre excited to build the data that athletes and coaches rely on. Experience at a sports data company is a big plus.Youre pragmatic. You pick the simplest approach that hits the bar, and you ship it.You own outcomes. When a model underperforms in the field, you dig in until its fixed.How Success Is MeasuredPrimary KPI: Accuracy of computer visionderived metrics against ground truth.Shipping: Models and improvements deployed to production on plan.Reliability: Consistent performance across the deployed camera fleet and real-world conditions.Efficiency: Inference latency and cost per game processed.QualificationsRequiredMS or PhD in Computer Science, Electrical Engineering, or a related field, with a focus on computer vision or machine learning.4+ years of industry experience in computer vision, including shipping models in commercial products.Strong Python and C++; deep experience with PyTorch (or similar) and OpenCV.Hands-on experience with object detection, multi-object tracking, camera calibration, and multi-view geometry or 3D reconstruction.Experience deploying and optimizing models in production (for example, ONNX, TensorRT, GPU optimization).Based in, or willing to relocate to, Austin or Los Angeles.PreferredExperience at a sports data or sports technology company (ball tracking, player tracking, or broadcast analytics).Background in baseball or softball, as a player or a deep fan of the game.Edge inference on embedded hardware (for example, NVIDIA Jetson).Pose estimation or biomechanics experience.Publications in computer vision or machine learning.
recblid kyuyy73b8s9uolz3a09dvdoprputmn