CoolFace
Apppublic

Vishal200/zomato-dispatcher-env

sourceHugging Facemitupdated 6mo agoView on Hugging Face
1likes
App README

Zomato Dispatcher OpenEnv

Food delivery order dispatch — a real-world RL environment built on the OpenEnv framework.

An AI agent acts as a fleet dispatcher for a food delivery platform inspired by Zomato/Swiggy. It observes incoming orders and available riders across city zones, then assigns orders to riders to maximise on-time delivery while avoiding SLA breaches.

Tasks

TaskDifficultyDescription
single_zoneEasy4 orders, 3 riders, 1 zone, generous deadlines
multi_zoneMedium10 orders, 4 riders, 5 zones, normal deadlines
peak_hourHard20 orders, 5 riders, tight deadlines + surge at step 10

Baseline scores (gpt-4o-mini agent)

TaskScore
single_zone1.00
multi_zone0.655
peak_hour0.475
Average0.71

Quick test

bash
curl https://Vishal200-zomato-dispatcher-env.hf.space/health
curl https://Vishal200-zomato-dispatcher-env.hf.space/tasks
curl https://Vishal200-zomato-dispatcher-env.hf.space/baseline

Endpoints

EndpointMethodDescription
/healthGETHealth check
/resetPOSTStart new episode
/stepPOSTExecute action
/stateGETCurrent state
/tasksGETList all tasks
/graderGETGrade episode
/baselineGETRun baseline on all tasks

Local setup

bash
pip install "openenv-core[core]>=0.2.1" fastapi uvicorn pydantic openai requests
uvicorn server.app:app --host 0.0.0.0 --port 7860

License

MIT