I’m building BetAssist, an AI-powered horse racing assistant that outputs: win/place probabilities a 0–100 runner rating three tailored bet suggestions per race (Safe, Smart, High-Roller) The mobile front-end is already designed. Your role is to build the entire backend system + modelling layer that powers these predictions. Framework & Data Needs The MVP will use a scikit-learn + gradient-boosted model (XGBoost / LightGBM / CatBoost). You will be responsible for: Data ingestion & freshness ingesting race-form archives collecting upcoming race data pulling live fixed-odds streams (TAB or equivalent) scraping or consuming public APIs keeping all data up to date daily Feature engineering Transform raw structured data into model-ready features such as: barrier bias class changes weight changes trainer/jockey strike rates form indicators distance/track suitability pace/early-speed proxies Key Deliverables ✔ 1. Reliable ETL pipelines Fetch, clean, store and update all required racing + odds data. ✔ 2. Training pipeline Reproducible environment (Docker preferred) that outputs: win/place probabilities AI runner rating (0–100) three bet suggestions per race Safe Bet Smart Bet High-Roller Bet 3. Production-ready API Token-secured REST API (FastAPI preferred) providing: race lists runner data predictions betting recommendations explanation text ✔ 4. Monitoring Simple checks for: data freshness API uptime model drift latency ✔ 5. Cloud deployment Deploy backend to AWS (EC2, ECS, or Lambda — your recommendation welcome). Must be ready for an MVP launch. Acceptance Criteria Your final API must: pull correct data for a held-out race day produce valid predictions for all runners deliver responses in <300 ms latency (warm) return complete JSON objects as documented be stable under basic load Milestone payments ONLY. What to Include in Your Proposal To be considered, please include: Relevant ML + backend projects (especially structured data, pipelines, APIs) Links to GitHub, portfolio or deployed apps Your approach to modelling & feature engineering Your proposed milestone structure Time estimate for full MVP Confirmation you’re comfortable with: scikit-learn + boosted-tree models milestone payments signing an NDA Ready to help me put this in punters’ pockets?