Polish & Finish Existing AI Fitness Tracker App (Pose Estimation + Routine Builder)

Description: I have a working AI fitness tracking app built with computer vision / pose estimation. The core functionality is already solid — it detects exercises in real time via webcam, tracks stages (up/down), counts reps, measures joint angles, and overlays everything on the video feed. The app is currently in a “testing code” state. I need someone to polish it, fix edge cases (especially ghost/false reps), expand the exercise library, and add a simple pre-workout routine builder so I can quickly select exercises before starting a session. This is a hobby/personal project, not a startup. I want it finished to a clean, usable level so I can actually rely on it for my own training. Current State: Core pose estimation and real-time exercise detection is working (example: Overhead Press detection + stage tracking) Rep counting exists but produces ghost/false reps while the model is acquiring the body or during unstable moments UI/overlay feels unfinished and cluttered Limited exercise support No pre-workout routine selection screen yet Required Deliverables: Overall Polish & UX Clean up the visual overlay (layout, text size/placement, colors, readability while lifting) Make the app feel finished and pleasant to use instead of “testing code” Pre-Workout Routine Builder (New Feature) Add a screen before tracking starts that displays buttons for available exercises User can tap/select multiple exercises to build a custom routine for that session Simple flow: select exercises → start workout → app tracks the chosen routine (with ability to switch exercises) Exercise Library Expansion Add reliable detection + stage tracking for additional common lifts Priority exercises to add: [replace this with your list, e.g. Squat, Bench Press, Deadlift, Bent Over Row, etc.] Edge Case & Robustness Fixes Significantly reduce or eliminate “ghost reps” (false positive rep counts while the model is still locking onto the body or during unstable/partial poses) Improve reliability in real home gym conditions (varying lighting, side angles, fatigue form, partial body visibility) Nice to Have (if budget/time allows): Basic form feedback / angle cues Easy highlight clip generation for content Simple session logging / history What I’m Looking For: Experience with computer vision / pose estimation projects (MediaPipe or similar is a plus) Portfolio examples of polished fitness, tracking, or camera-based apps Ability to work with an existing codebase and improve it without over-engineering Deliverables & Milestones: Milestone 1: Polish + ghost rep reduction + basic routine builder screen (core usability) Milestone 2: Exercise expansion + full robustness testing Final: Clean, usable app that I can run for real workouts Please include in your proposal: Relevant experience with similar computer vision or fitness tracking projects Estimated timeline Fixed price or hourly rate with estimated hours