# GreenOps

> Carbon-aware scheduling for data warehouse jobs: waits out dirty hours, still meets deadlines.

A project by [Jo Saillier](https://joransaillier.com/), AI engineer. Category: optimisation. Web page: https://joransaillier.com/work/greenops

## Facts

- Team: Jo and two mathematicians
- Event: TEB AI Hackathon, Vienna
- When: 23 to 25 Oct 2026

## In short

GreenOps is a carbon-aware cloud scheduler built for Exoscale's GreenOps challenge: 15 cloud jobs, 11 zones, 120 simulated hours. It pauses flexible jobs through high-carbon hours and restarts them in cleaner ones, and every job must still meet its deadline. All grid data is simulated.

Scheduling decisions have to be fast and repeatable, so Jev, a classification model, picks in real time among options that code has already built and checked. Code validates every answer, so each decision stays deterministic. That speed made the team the first to complete all 15 workloads within every constraint.

Built by a team of three at the TEB AI Hackathon in Vienna, with two mathematicians on forecasting and energy estimates. Every run was scored on the same KPIs and carefully monitored against overfitting. The best run delivered all 15 jobs on time with 8.7% less electricity CO₂e than an earlier run.

## Stack used

- Controller: Python standard library, read-only Docker container, CarbonScaler method
- Simulator and model: Exoscale simulator, grid carbon forecasts, Jev classification model

## Contact

About GreenOps or a similar project: https://joransaillier.com/#contact · LinkedIn: https://www.linkedin.com/in/joran-saillier/
All work: https://joransaillier.com/
