App Prediction Bot Development

I need a self-contained prediction bot that plugs straight into my existing mobile app and delivers accurate forecasts in real time. The core job is to design the model, build the backend service, and expose everything through an API so my front-end team can call a single endpoint and display the result. Here’s what I have and what I expect: • Existing app codebase (React Native) with user authentication and a dedicated screen waiting for the prediction output. • Flexible data sources: I can supply anonymised in-app data and I’m open to bringing in third-party APIs if your approach needs them. • Clear, concise documentation outlining the model architecture, data flow, and any setup steps so my devs can re-deploy without your help. • A simple test harness or Postman collection that proves the endpoint works and returns consistent predictions. I’m open on tech stack but favour Python (FastAPI), Node.js, or anything you can containerise easily in Docker and deploy to AWS. If a lightweight model in TensorFlow, PyTorch, scikit-learn, or even a rules-based engine fits the brief, I’m happy to go with what produces the best accuracy-to-latency balance. Acceptance criteria 1. A single REST or GraphQL endpoint returning a prediction and confidence score. 2. End-to-end latency under two seconds on a small AWS t3.medium instance. 3. 90 % test coverage on core logic and a short README that lets my team run “docker compose up” and see it live. If you have a different idea that achieves the same goals faster or more reliably, let me know in your proposal.