Replike/RepDetect is a web application that uses real-time pose detection with MediaPipe to count your exercise repetitions. It is a Next.js 14 + TypeScript app currently deployed at replike.vercel.app. Replike does all processing locally and the …
Replike/RepDetect is a web application that uses real-time pose detection with MediaPipe to count your exercise repetitions. It is a Next.js 14 + TypeScript app currently deployed at replike.vercel.app. Replike does all processing locally and the data lives in localStorage. It is a working MVP with pretty nice detection logic however it needs a few important features and fixes I hope on delivering in the future. But a reminder that the features currently present are very nice to use and grasp the core idea of this project, give it a go!
I was responsible for the core concept, feature set, and overall UI/UX design of this project. I defined how the app should work, what problems it solves, and how users interact with it.
I used AI tools to assist with implementing parts of the code, since I am still developing my programming skills. However, this was not a one-shot generation process — I iterated extensively on the outputs by refining prompts, modifying logic, fixing issues, and testing each feature to make sure it behaved as intended.
I made decisions about how the system should function, how features should be structured, and how the user experience should feel. I also tested the application myself, debugged problems, and handled deployment.
Throughout the project, I focused on understanding the code and how each part works rather than blindly using generated output. AI was used as a tool to support development, while the direction, design, and iteration process were driven by me.