A curated collection of research, tools, and best practices for environmentally sustainable AI development and deployment.

Eco CI Energy estimation for Github Actions Runner VMs

Measure energy and carbon consumption of software

Mock Ada Carbon Monitoring Implementation

A neat helper that manages the standby state of unix hosts with Wake-On-Lan configured, with Web-GUI.

PowerAPI is a Python framework for building software-defined power meters.

SmartWatts is a formula for a self-adaptive software-defined power meter based on the PowerAPI framework.

A Prometheus exporter and a REST API server to export metrics of compute units of resource managers like SLURM, Openstack, k8s, _etc_

An open-source architecture for AI data centers that use zero freshwater. This repo provides practical designs for replacing evaporative cooling with closed-loop immersion, heat-to-power recovery, and adaptive AI-based thermal control, reducing water usage by 100% and energy demand by up to 20%.Built through a cross-AI collaboration.

⚡ Blazing fast Git hooks manager written in Rust. Drop-in replacement for Husky with 27x faster startup. Zero dependencies, parallel execution, and built-in carbon footprint tracking.

Monitoring power consumption and calculates CO2 emissions from Kubernetes containers

Carbon-aware scheduling and traffic management Kubernetes Operator. It dynamically delays and redirects requests based on real-time energy mix data, optimizing workloads for greener energy consumption.

A plugin named wall-e-impact that uses the Impact Framework to measure game applications and their carbon emissions.

A central place to collect and organize studies on digital sustainability.

Green Software Use on HPC

a Kubernetes operator that aggregates energy and CO2 emission data for tagged resources

[FGCS] Code and data for the paper "Adaptive green cloud applications: Balancing emissions, revenue, and user experience through approximate computing"

The Green Computing Scheduler is a carbon-aware Kubernetes scheduler that helps organizations reduce their carbon footprint by intelligently scheduling workloads based on real-time carbon intensity data. By delaying non-urgent batch jobs until periods of lower carbon intensity, the scheduler can significantly reduce the carbon emissions associated