AI Exercise Form Analyzer

I’m building a mobile app that gives real-time, AI-driven feedback on exercise technique. The first release will focus on classic strength-training moves—think squats, deadlifts, presses—before we branch into areas such as yoga. I already have the cloud and mobile infrastructure in place; what I need is a partner who can chart the technical course, train the models, and wire everything together so that end users see instant coaching cues on their phones.

Here’s how I picture the collaboration:

• Direction & Road-mapping – Help me choose the optimal computer-vision approach, keypoint tracking libraries, and model architecture so the system recognizes each rep accurately under varied lighting and camera angles.
• Data & Model Work – Curate or generate a robust training set, build the pose-estimation and classification models, and iterate until the app reliably flags depth, knee alignment, back angle, and tempo mistakes.
• App Integration – Embed the inference pipeline into the existing iOS/Android codebase, keeping latency low enough for live feedback without noticeable lag.
• Feedback Logic – Design the rule set that turns raw joint data into user-friendly corrections (audio prompts, on-screen overlays, rep scores).
• Validation – Provide objective accuracy metrics and a simple test protocol so I can verify performance on new footage.

Please send a detailed project proposal outlining your technical plan, timeline, and any similar form-analysis work you’ve shipped before. I’m ready to start as soon as we agree on the approach.