Control carbon emission of your cloud infrastructure - Carbon-aware cloud infra
-
Updated
Jul 13, 2024 - Go
Control carbon emission of your cloud infrastructure - Carbon-aware cloud infra
A Python library for simulating datacenter energy systems ⚡️
The "Economy Mode" for AI Compute
Simulator and datasets to research on carbon-aware temporal workload shifting.
[ACM e-Energy'24] Code and data for "FedZero: Leveraging Renewable Excess Energy in Federated Learning"
A carbon-aware ML training scheduling system that optimises machine learning workloads using real-time UK grid carbon intensity data. It simulates ML training, finds optimal low-carbon execution windows, and visualises carbon savings through an interactive multi-page Streamlit analytics dashboard.
a Kubernetes operator that aggregates energy and CO2 emission data for tagged resources
API d'intensité carbone de l'électricité française (gCO₂eq/kWh) — souveraine, open source, dev-first. L'équivalent FR de carbonintensity.org.uk, sur données ouvertes RTE/ODRÉ 🇫🇷.
This tool provides a real-time and historical perspective on the carbon intensity of electricity consumed within Great Britain, offering academic and industry insights.
Privacy-preserving, adaptive mechanism that learns per user/domain preferences for sustainability without degrading search quality.
Carbon-aware load scheduler: shifts EV charging, water heating, laundry and pumps into the cleanest grid hours without missing a deadline. Next.js + FastAPI + OR-Tools CP-SAT, live weather-based carbon signal, System One decisions, execution tracking.
Run non-urgent LLM tasks at the cleanest grid hour before a deadline. MCP server + TS/Python library. Apache-2.0.
Carbon aware suspension for stateful Apache Flink jobs on Kubernetes, without losing their state
Predict the energy consumption of Kubernetes workloads from their specs, before they run. Ground-truth energy labels from Kepler (eBPF) via Prometheus.
Demand-aware replica scaling and placement for microservices in geo-distributed Kubernetes federations. A central control plane forecasts the total demand, sets the replica budget and places replicas by reachable demand, weighing latency, capacity, network proximity and carbon intensity. Code, campaign and analysis of DMOS.
A simulation game for teaching Green IT – from code to data center. Run a data center and improve its software so that it serves its users reliably and emits as little CO₂ as possible. Web and desktop, English and German.
Open-source PyTorch AI model profiler for latency, memory, energy (NVML/RAPL), and layer-wise bottlenecks. Measure green AI / ML efficiency no fabricated joules.
Add energy/carbon/resource-cost awareness to software via Eclexia
🌍 TypeScript superset for SDGs-aware programming. Write sustainable code aligned with UN Sustainable Development Goals through static analysis, runtime monitoring, and developer tools.
Carbon-aware cloud scheduler run live on 13 AWS regions, with 25k verified forecasts and a guard that disables models that don't deliver
To associate your repository with the carbon-aware topic, visit your repo's landing page and select "manage topics."