GelAl Autopilot
Foodathon by ATP CapitalApr 2026Data & ML engineer (team of 4)
A product that nudges food-delivery users toward pick-up orders with a propensity model and a recommendation engine that sets the minimum personalized discount.
Highlights
- Engineered 37 behavioral and geospatial features from 1M orders and 100K users
- Trained gradient-boosted propensity models on 441K orders
- Served recommendations with a rule-based fallback through FastAPI
