Hacking Sustainability

These projects were completed by the class of 2026

Group 0: Template Project
By Student1 first and last name, Student2, Student3.
This is a summary with a max of 200 characters; The links below should send the reader to your paper, the tool you folks built (source code or website), and optionally your presentation video. Please remove yaml entries for links you do not use.
Paper. Website. Source code. Video Presentation.

Group 1: GCI - Wasteful CI Analyzer
By Jayran Duggins, Priyansh Rajusth, Erkin Başol, Maja Bińkowska, Nicolas Loaiza Atehortua.
Continuous Integration (CI) pipelines automatically build and test software when changes are made to a repository. While CI improves software reliability, many runs provide little value while still consuming compute resources and energy. Examples include rerunning jobs after flaky test failures, scheduled workflows running on inactive repositori…
Paper.Source code.

Group 2: EcoCode
By Aiman Abdul Wahab, Carolyn Alcaraz, Ceylin Ece, Gönenç Turanlı, Miguel El Khal.
EcoCode is a command-line tool that helps developers choose the most energy-efficient AI model for code generation. Users provide a coding prompt through the CLI, and EcoCode analyzes its complexity to recommend an appropriate model tier (e.g., small or large). The tool then displays key information such as the recommended model, estimated token…
Paper. Website. Source code.

Group 3: Patching software inefficiencies
By Calin Georgescu, Nick van Luijk, Vincent Ruijgrok, Radu Andrei Vasile, Arnas Venskūnas.
This paper explores a novel set of guidelines to practically apply energy-saving optimizations at runtime for desktop applications. Currently, many tactics exist to optimize energy efficiency, but a central, practical, and industry-oriented set of guidelines is lacking. To address this gap, we conduct an investigation of prior research describin…
Paper.Source code.

Group 4: Sustainable Certification website
By Moniek Tummers, Anouck Heutinck, Maksym Ziemlewski, Alexandru Marin, Vasil Chirov.
For this project we developed a web-based tool that centralises knowledge about sustainable software certifications. The application gathers existing certifications and provides structured summaries, making information more accessible and easier to navigate. Furthermore, it includes a questionnaire that helps organisations and other individuals …
Paper. Website. Source code. Video Presentation.

Group 5: Carbon Aware Thermostat
By Robin Kruijf, Piotr Kranendonk, Anhar Al Haydar, Pranav Pisupati, Jari de Keijzer and Emīls Dzintars.
For project 2 we want to implement a carbon aware thermostat controller for Home Assistant (https://www.home-assistant.io). The controller should shift the temperature based on the current energy grid (how clean it is: solar energy, wind energy or fossil fuel). The controller will use carbon intensity data, weath…
Paper.Source code.

Group 6: GreenTips: A Sustainability Tip a Day Makes Energy Waste Go Away
By Wojciech Mundala, Ada Turgut, Sahana Ganesh, Valantis Andreas, Ana Mako.
GreenTips is a tool that enables developers to explore alternative sustainable design patterns by giving information in the form of tips during coding. It aims to optimize code in order to save energy consumption. It essentially promotes sustainable coding practices and increases sustainability awareness for developers in a format that is non-in…
Paper. Website. Source code. Video Presentation.

Group 7: Nuclear Mix for Code Carbon
By Roham Koohestani, Kunal Narwani, Nina Semjanová, Caio Miranda Haschelevici, Georgios Markozanis.
This paper presents NuclearMix, a CodeCarbon extension supporting lifecycle-aware electricity-mix accounting, alongside a scrollytelling website on nuclear emissions. A paired pre/post survey (N=17) showed statistically significant knowledge gains (Cohen’s d=1.08). Technical analysis revealed that reported software emissions vary substantially d…
Paper. Website. Source code. Video Presentation.

Group 9: GreenLint: Energy-Aware Code Analyzer for Python
By Maciej Wierzbicki, Mikołaj Magiera, Figen Ulusal, Radu Chiriac, Ibrahim Badr.
GreenLint is a lightweight Python linter that detects energy-inefficient code patterns using static AST analysis. The tool focuses on common inefficiencies such as suboptimal loops, repeated operations, and non-vectorized data processing. Unlike existing tools that estimate overall energy consumption or focus on general code quality, GreenLint p…
Paper.Source code.

Group 10: A Linter and Guideline Framework for Sustainable GitHub Actions Workflows
By Rebecca Andrei, Boris Annink, Paul Anton, Kasper van Maasdam, Radu Serban.
We created a CI/CD rulebook and workflow linter for GitHub Actions named ‘suslint’ to assist developers in creating more sustainable workflows.
Paper. Website. Source code.

Group 11: What is killing my computer? Measuring and recording background energy consumption
By Miroslav atanasov, Andreas Tsatsanis, Francisco Duque De Morais Amaro, Raluca Alexia Neatu, Thomas van der Boon.
While many energy profiling tools and methods are out there, there is little incentive for application developers to spend their time optimising for energy consumption. This widespread practice takes a toll on consumer hardware, and can result in premature end-of-life of otherwise capable devices. Our goal is to provide a lightweight tool to mon…
Paper. Website. Source code. Video Presentation.

Group 12: Energy-efficient programming AI
By Fedor Baryshnikov, Yuting Dong, Yuchen Sun, Tobias Veselka, Tom Clark.
Many programmers use AI tools in their workflow, but often use computationally and thus energy intensive models for simple tasks, wasting energy. Our tool is an LLM interface that programmers can use for their needs - it is similar to existing solutions meaning it is easy to get used to, but uses a novel Planner-Worker approach to save energy wi…
Paper.Source code.

Group 13: Flow-Sate: Developer Well-being Plugin
By Konstantina Anastasiadou, Zofia Rogacka-Trojak, Amy van der Meijden, Andriana Tzanidou, Jimmy Oei.
We present Flow-State, a VS Code IDE plugin designed to enhance developer productivity and flow, reduce cognitive load, and promote well-being. Our design draws inspiration from the SPACE and DevEx frameworks. The plugin consists of various features for monitoring individual cognitive load, such as the add-to-delete ratio and large code insertio…
Paper.Source code.

Group 14: GreenField — Cross-Boundary JSON Field Analysis
By Preethika Ajaykumar, Atharva Dagaonkar, Riya Gupta, Sneha Prashanth, Deon Saji.
Modern full-stack applications silently accumulate unused JSON fields, data the backend sends but the frontend never reads, or fields the frontend submits but the backend never processes. This unused payload wastes CPU cycles on serialisation, consumes unnecessary network bandwidth, and drains mobile battery at every request, yet no existing too…
Paper.Source code. Video Presentation.

Group 15: Carbon-Aware Scheduling
By Arda Duyum, Yuvraj Singh Pathania, Brewen Couaran, Taeyong Kwon, Elia Jabbour.
A carbon-aware scheduler that shifts batch AI jobs like training and inference to time windows with the cleanest electricity grid, leveraging real-time data from existing databases.
Paper. Website. Source code.

Group 16: GreenestRoute: Carbon-Aware Job Scheduling for a Greener ICT Sector
By Yanzhi Chen, Alex Hautelman, Daniel Rugge, Sydney Kho, Yi Wu.
Software can be run almost anywhere and at any time, yet this flexibility is under-exploited for cutting carbon emissions. GreenestRoute is an open-source scheduler that routes flexible compute jobs to a server location and start time that jointly minimise CO2 emissions and server cost, making green scheduling a practical default. It integrates …
Paper.Source code.

Group 16: Github Actions to report the energy consumption of CI/CD
By Emre Cebi, Alexandru Mitteltu, Kevin Shan, Anton Cosmins, Mohammed Nassiri.
An application that integrates with GitHub Actions to track energy consumption and provide developers with actionable insights on their sustainability impact.
Paper.Source code. Video Presentation.

Group 17: GDPR Compliance Learning Tool
By Antonio Lupu, Diana Sutac, Maria Cristescu, Jeroen Chu, Sophie (Schaaf) Langeveld.
We propose a tool that developers can use to learn about GDPR and check if they are compliant with it.
Paper. Website. Source code. Video Presentation.

Group 18: LeafCode
By Ayush Khadka, Konstantinos Syrros, Medon Abraham, Norah E. Milanesi, Job Stouthart.
As the environmental impact of computing scales, traditional algorithmic platforms remain focused on time-memory optimization, neglecting energy efficiency. Various developers do not have the necessary skills to implement energy efficient code.We present LeafCode, a gamified learning platform designed to foster Green Software Engineering (GSE) p…
Paper. Website. Source code.

Group 19: EcoTracker: An Application-Level Carbon Scheduling Framework for LLMs
By Adomas Bagdonas, Georgi Dimitrov, Kristian Hristov, Stilyan Penchev, Daniel Rachev.
The rapid growth of Large Language Models has intensified concerns about the environmental cost of AI inference. Currently, developers using third-party LLM APIs lack plug-and-play solutions to actively mitigate their footprint. To address this, we developed EcoTracker, an open-source Python framework published on PyPI for application-level carb…
Paper. Website. Source code. Video Presentation.

Group 20: Chrome Extension: Sustainable Mode
By Ignas Vasiliauskas, Jeffrey Meerovici, Nicolas Hornea, Benas Pranauskas, Arnav Biswas.
Browsing might use more energy than a user imagines, on the browser device itself and on the datacenters running applications. We want to implement a browser extension that would allow to enable sustainable mode. In this mode, users would be able to limit scripts running on websites, disable autoplay, automatically run videos in lower quality, d…
Paper. Website. Source code.

Group 21: Group 21: Jamanota middleware for tracking energy usage in Langchain
By Andrea Vezzuto, Jan Kuhta, Rodrigo Montero Gonzalez, Aadesh Ramai, Samuel van den Houten.
The rapid growth of AI agent systems in recent years has introduced significant environmental concerns. However, developers currently lack the tools to assess the energy and carbon footprint of their applications. Existing observability platforms expose token usage as a proxy for cost, but do not translate this into meaningful environmental metr…
Paper. Website. Source code.

Group 22: Dockerfile Carbon Optimizer
By Ion Tulei, Alexandru Verhovetchi, Horia Zaharia, Dragos Erhan, Joost Weerheim.
A CLI tool that analyzes Dockerfiles for energy-wasteful patterns, estimates carbon cost per finding, and auto-fixes them.
Paper.Source code.

Group 23: FAST: Energy-Aware Test Prioritization for Fail-Early Software Testing
By Dibyendu Gupta, Cem Gungor, Uddhav Pisharody, Noky Soekarman, Leonidas Hadjiyiannis.
FAST (FAil early in Software Testing), a VSCode extension for JavaScript projects that measures the energy consumption of individual test cases and prioritizes them accordingly. The extension profiles existing tests, ranks them in ascending order of measured energy use, and executes them stopping at the first failure.
Paper. Website. Source code. Video Presentation.

Group 24: Flow: a Carbon-Aware CI/CD Scheduler
By Cristian Benghe, Antoni Nowakowski, Andrei Paduraru, Poyraz Temiz, Tess Hobbes.
Flow is a carbon-aware CI/CD scheduler that defers non-critical GitHub Actions workflows to periods of low grid carbon intensity, achieving up to 80.9% carbon reduction.
Paper. Website. Source code.

Group 25: Carbon Clicker
By Conall Lynch, Arjun Rajesh Nair, Viktor Shapchev, Coen Werre, Noah Tjoen & Gabriel Leite Savegnago.
Our project was the creation of ‘Carbon Clicker’ An Engaging Learning tool about Greenwashing in the Tech Sector.
Paper. Website. Source code.

Group 26: Towards LLM Energy Labels: Accuracy and Energy Efficiency
By Levi Ari Pronk, Ocean Wang, Nicholas Wu, Madhav Chawla, Yasar Saltuk Bugra Kocdas.
We propose Energy Per Correct Answer (EPCA), a metric combining energy cost and accuracy, and a proof-of-concept tool that benchmarks LLMs across coding, math, and logical reasoning domains.
Paper.Source code.

Group 30: JouleDuel: Energy-Aware Programming Platform
By Alessandro Valmori, Bill Vi, Wilhelm Marcu, Maciej Bober, Frederik van der Els.
We present JouleDuel, an energy-aware competitive programming platform designed to foster sustainable software engineering practices through gamified challenges.
Paper. Website. 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 2026/p2_hacking_sustainability
    • Use the following filename format: g<group_number>_<1/2meaningful_keywords>.md
    • Use the file gX_template.md as a template
    • If you want to add images, add it to 2026/p2_hacking_sustainability/img/g<group_number>_<1/2meaningful_keywords>/
  3. Commit, Push.
  4. Submit a pull request. Make sure to use the pull request template for project 2.

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.

If you are looking for a paper template to write your report in, you could take the IEEE conference paper 2-column template: https://www.overleaf.com/latex/templates/ieee-bare-demo-template-for-conferences/ypypvwjmvtdf

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/2026/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.