Hacking Sustainability

Group : Greenchat
By Simon Biennier, Jasper Heijne, Paul Lindhorst and Huib Sprangers.
This paper introduces GreenChat, a browser extension designed to provide real-time feedback on the environmental impact of ChatGPT conversations. By estimating energy consumption, carbon emissions and water usage, the tool aims to increase user awareness and encourage more sustainable AI usage.
Paper. Website. Source code.

Group : Energy-Efficient ML Workloads in Docker: Measurement and Evaluation using EnergiBridge
By Ahmed Driouech, Ahmed Ibrahim, Taoufik el Kadi, Moegiez Bhatti.
This project measures energy use of ML workloads in Docker, showing CPU-optimized containers reduce power consumption across models and OSs.
Paper.Source code.

Group 1: GreenCodeAnalyzer: Detecting Energy Code Smells in Data Science with Static Analysis
By Marina Escribano Esteban, Kevin Hoxha, Inaesh Joshi, Todor Mladenovic.
As data science workloads continue to grow in complexity and scale, energy efficiency has become a critical concern. Code inefficiencies not only inflate computational costs but also exacerbate environmental impacts. While existing tools profile and measure energy consumption of code at runtime or statically analyze energy code smells, few solut…
Paper. Website. Source code.

Group 2: Energy Efficiency in LLM Inference: Comparing Inference Libraries in a Unified Docker Framework
By Reinier Schep, Razvan Loghin, Maosheng Jiang, Alex Zheng.
The increasing adoption of Large Language Models (LLMs) has raised concerns about their computational efficiency and energy consumption. This study presents a comparative analysis of four popular LLM inference libraries—Ollama, MLC, vLLM, and TensorRT—evaluating their energy efficiency in a standardized Dockerized environment. Each library is te…
Paper. Website. Source code.

Group 3: Energy-Aware Software: Intelligent Scheduling of Energy-Intensive tasks with Grid Congestion Data
By Matthijs Vossen, Melle Koper, Roan Rosema, Scott Jochems.
This project addresses the integration of grid congestion awareness into software systems to enhance sustainability and grid efficiency. We highlight the importance of accurate load forecasting in preventing energy waste, avoiding renewable energy curtailment, and optimizing electricity use during surplus production periods. The solution extends…
Paper. Website. Source code.

Group 4: Green Llama: A Tool for Monitoring Energy Consumption and Sustainability in Local LLMs
By Anyan Huang, Philippe A. Henry, Yongcheng Huang, Yiming Chen.
Large language models (LLMs) have revolutionized artificial intelligence by delivering unprecedented performance in various applications. However, their substantial computational demands have raised critical concerns regarding high energy consumption and environmental impact, especially during infer- ence on local machines. This paper introduces…
Paper.Source code.

Group 5: Finding energy hotspots in JavaScript using ESLint
By Jort van Driel, Dorian Erhan, Weicheng Hu, Giannos Rekkas.
This study examines how everyday JavaScript coding decisions impact a web application’s energy use. We introduce ESLint-based rules that detect “green” or “inefficient” patterns and then compare power consumption for each variant by measuring actual CPU energy. Twelve rules were tested, examples include using lazy image loading, preloading asset…
Paper. Website. Source code.

Group 6: Energy Consumption of Static Analysis Tools
By Rafał Owczarski, Lászlo Roovers, Athanasios Christopoulos, Muhammad Zain Fazal.
As software development increasingly emphasizes sustainability, understanding the energy consumption of development tools has become essential. Static analysis tools, such as PMD, help improve code quality by detecting issues early in the development process. However, their computational cost and energy footprint remain largely unexplored. This …
Paper.Source code.

Group 8: Hugging Carbon: Visualising the Carbon Emissions of Hugging Face’s AI Models
By Kevin Yi Chen, Tiberiu Sabău, Floris van Veen, Kah Ming Wong.
Website created to educate users on the carbon emissions of huggingface’s ai models. Users would be able to visualise the performance, carbon emissions and are given suggestions on better models to use.
Paper. Website. Source code.

Group 9: Dataset of government-developed OS software
By Ilma Jaganjac, Angelos-Ermis Mangos, Marvin Blommestijn, Pravesha Ramsundersingh.
(Visit our website!) This project creates a dataset and analysis framework to evaluate the sustainability of government-developed open-source software. It collects repositories from five countries: the United States, the Netherlands, Greece, Germany, and France, and assesses them across five dimensions: technical, environmental, economical, soci…
Paper. Website. Source code.

Group 10: Measuring Energy Consumption of Different Security Static Analysis Tools and Configurations
By Andrea Onofrei, Ayush Kuruvilla, Sahar Marossi, Yulin Chen.
This study introduces a reproducible framework for profiling the energy consumption of security-focused static analysis tools—Bandit and Semgrep—across varying configurations and codebases. By measuring CPU energy usage on three Python projects of increasing complexity, the research highlights how tool architecture and project size influence ene…
Paper. Website. Source code.

Group 11: Measuring the Energy Consumption of Docker Images for ML Workloads
By Ana Țerna, Andrei-Iulian Vișoiu, Lucian Toșa, Monica Păun.
Investigation on how Docker image configurations affect consumption during inference in ML workloads. We use a ResNet-50 image classification model on randomly generated images of various resolutions.
Paper.Source code.

Group 13: Enhancing EnergiBridge: A Service-Based Approach to Energy Profiling
By Sofia Konovalova (6174019), Kaijen Lee (5100887), Violeta Macsim (5498031), Jakub Patałuch (5514274).
EnergiBridge, an open-source energy profiler, is now service-based for better integration. Implemented in Rust and C++, it enables function-level monitoring via JSON-RPC. They enable function-level monitoring with minimal overhead, ensuring accurate energy measurement.
Paper.Source code.

Group 14: Measuring Energy Consumption of JUnit Tests
By Joaquín Cava, Elena Ibañez, Roelof van der Geest, Jeroen Janssen.
Unit testing is a fundamental part of software development. We measured energy usage of various JUnit test suites belonging to open-source Java projects. Using our own experiment pipeline we insert JoularJX into the project definitions to enable measurement of individual unittests. With the results we were able to define a few test definitions t…
Paper.Source code.

Group 15: ChatGPT Carbon Tracker
By Raghav Talwar, Konrad Barbers, Peiyan Liu, Shalakha R S.
The ChatGPT Carbon Tracker project focuses on the often-ignored environmental effects of large language models during user interactions. It does this by creating a lightweight browser extension that runs on the client side, which estimates and shows the carbon footprint of using ChatGPT in real time. By counting the tokens in messages from both …
Paper. Website. Source code.

Group 16: Friction and Tiredness: A literature analysis of burnout causes in the field of software engineering
By Gopal-Raj Panchu, Mirko Boon, Sotiris Vacanas, Reeve Lorena.
In this research paper we analyze existing literature regarding burnout in the software engineering field.
Paper.Source code.

Group 18: Estimating and Visualizing Energy Consumption of LLMs within IDEs: A VS Code Extension for GitHub Copilot
By Gyum Cho, Cyprian Bîcă, Denis Krylov, Matteo Fregonara.
A VS Code extension that reveals the energy and CO₂ cost of AI code suggestions in real time, promoting transparency and sustainability in LLM-assisted development.
Paper. Website. Source code.

Group 20: HowSUS: A Sustainability Scoring Framework for Open-Source Libraries
By Seyidali Bulut, Johan van den Berg, Michal Kuchar, Artin Sanaye.
This project introduces HowSUS, a framework to assess the sustainability of software libraries using metrics such as performance, maintainability, community, and security.
Paper. Website. Source code.

Group 21: Sustainable Sprint
By Martijn Frericks, Wout Burgers, Thomas Rooskens, David van der Maas.
Sustainable Sprint is an educational board game inspired by ‘The Game of Goose’ designed to teach sustainable software development principles in an engaging way. Recognizing that the environmental impact of software is often overlooked, the game aims to make developers aware of how inefficient coding contributes to energy waste. Players navigate…
Paper. Website. Source code.

Group 22: GradlEnergi: Raising awareness about local build pipeline energy consumption
By Gijs Margadant, Jamila Seyidova, Michael Chan, Roberto Negro.
In this paper we present a novel tool to aid in measuring energy consumption of specific tasks in gradle build pipelines
Paper. Website. Source code.

Group 23: Evaluating Power Consumption of Python Test Generation using Pynguin
By Victor Hornet, Elena Mihalache, Andreea Mocanu, Alexandru Postu, Kian Sie.
The report examines Pynguin’s energy use in test generation. MIO uses more power but is faster, while MOSA and DynaMOSA use less power but take longer, offering guidelines for sustainable testing.
Paper. Website. Source code.

Group 24: Cross-Machine Comparable Benchmark for Machine Learning Energy Consumption
By Luc Dop, Nawmi Nujhat, Sabina Gradinariu, Vincent van Vliet.
This paper introduces a machine-agnostic benchmark for measuring and comparing the energy consumption of machine learning models during inference. Built on top of MLPerf and extended with energy tracking tools like EnergiBridge and CarbonTracker, the benchmark enables fair cross-machine evaluations. Using EfficientNet and Vision Transformer on t…
Paper.Source code.





How to contribute

To add a new article, follow the instructions below:

  1. Fork the repo of the website on Github: https://github.com/luiscruz/course_sustainableSE/
  2. Create a new markdown file inside the directory 2025/p2_hacking_sustainability
    • Use the following filename format: g<group_number>_<1/2meaningful_keywords>.md
    • Use the file gX_template.md has a template
    • If you want to add images, add it to 2025/img/p2_hacking_sustainability/g<group_number>_<1/2meaningful_keywords>/
  3. Commit, Push.
  4. Submit a pull request.

Explaining the template. Although it is a markdown (.md) file, you will only be filling the YAML header with some keys and values. In particular, you must fill author, title, summary with a quick description of the project (max 200 characters), and paper with a url link to the paper. Optionally, you can also fill image with the url of a logo or image related to the project, source with a link to the source code of the project, and website with a link to the project’s website when applicable.

Before submitting the pull request, you should test whether your file is rendering properly in the website. The easiest way to check it is by running the docker container, as instructed in the Github Readme.

Your page should be listed here: http://localhost:4000/course_sustainableSE/2025/p2_hacking_sustainability

If you don’t want to deal with jekyll, you can do it the slow and expensive way: 1) enable github pages in your fork repo 2) check your the deployed page. (I don’t recommend it, though)

Note: let me know if you run into any issue or if there’s any step you think should be explained here.