Computer Vision
Build the models that turn game film into tracking, events, stats and searchable moments.
About AirPLAi Sports
AirPLAi is building AI video intelligence for sports. We turn raw footage from any camera into customized, professional-level insights from every game. Our platform is live with marquee partners spanning professional basketball, elite youth and developmental leagues, women's sports, and leading sports media brands.
The models are the product. Every stat, highlight, coaching breakdown, and profile we deliver starts with vision systems that understand what happened on the court. We are expanding from basketball to a multi-sport platform, and the vision stack is where that expansion happens.
The Role
We are hiring a Computer Vision Engineer to build and own the models that turn game film into a structured game record: player detection and tracking, court and field registration, event and action recognition, and player identification.
You will work across the full model lifecycle: curating and labeling data from real games, training and evaluating models, optimizing inference for cost and latency, and watching your work show up in live products days later.
This is an early hire with real ownership: your calls on architecture, data, and evaluation set the standard for how AirPLAi does vision as we scale across sports.
What We're Looking For
Must have
- Production computer vision experience: you have shipped detection, tracking, or video-understanding models that real users depended on.
- Strong PyTorch fundamentals and comfort taking models from notebook to optimized inference (batching, quantization, GPU serving).
- Classic vision fundamentals: camera calibration, homography and projective geometry, multi-object tracking.
- Evaluation discipline. You build eval sets and metrics before you trust a model, and you know how to find where it fails.
- Experience working with messy real-world video: varied camera angles, lighting, occlusion, and broadcast or phone footage.
- You have owned a vision system end to end somewhere before, and you shipped.
The X-Factor: You watch game film for fun. Knowing what a pindown or a backdoor cut looks like makes you better at this job.
What You'll Do
You will own the pipeline from frames to structured game events: detection and multi-object tracking tuned for sports, homography that maps any camera angle onto the court, and event models that turn possessions into stats, shot locations, and searchable moments. You will build the training and evaluation loops that let us improve from every game we process, and the inference paths that keep live-game latency and GPU cost in check.
You will also drive the multi-sport expansion: adapting the stack to new court and field geometries, new event vocabularies, and new footage styles, working directly with product and infrastructure. Engineers at AirPLAi have product influence, not just a ticket queue.
Compensation
Competitive cash compensation and meaningful equity, based on experience and role structure. We are open to discussing the right structure for an exceptional candidate.
Prefer Email?
Apply with the form, or email Amaan directly at amaan@airplaisports.com with your resume or LinkedIn profile. Tell us about a vision system you shipped, and what fooled it.
Not sure this exact role is the right fit? Reach out anyway. We are growing fast, have no shortage of hard problems to solve, and will have other roles opening soon. We care far more about what you have built than a checklist.