ML Hackathon

Data is the new oil, but it takes a truly brilliant engineer to refine it. Welcome to the Machine Learning Hackathon of Fugacity 2026, the ultimate battleground where data science meets chemical engineering! Modern chemical plants generate millions of data every second from various sensors measuring temperatures, pressures, flows and compositions. You’ll be working on industrial datasets to build predictive models to solve real-world chemical engineering problems. So go ahead, tame the chaotic sensor logs before they wreck the plant! PS release date: 20th July  Submission Round Ends: August 16, 2026 Result Declaration of Finalists: August 19, 2026 Grand Finale: August 30, 2026 Venue of the Finale: Department of Chemical Engineering, IIT Kharagpur Event Format: Round 1—Online Qualifier: Participants will be given access to a curated dataset and a problem statement. Teams must build and train their models, and submit the prediction file. Round 2—Offline Presentation: The highest ranking teams on the leaderboard will be presenting their solutions to the jury at the Department of Chemical Engineering, IIT Kharagpur. Guidelines Eligibility: All undergraduate and postgraduate engineering students across India Team Configuration: 1 to 4 members. Cross-departmental and inter-college teams are highly encouraged, but a participant can only belong to a single team.  Final submissions must include the predictions in the file format that’s specified in the problem statement. Finalists are also required to submit fully documented .ipynb files. The finalists will then be judged based on their innovation, robustness and scalability of their model and their pitch. Rules Your code must run end-to-end without errors. All random numbers and data splits must be explicitly set so that the judges can replicate your exact scores. Submissions will be scored instantly using the metrics defined in the problem statement. Finalists will be declared based on the leaderboard. Your notebook must be with clean code blocks and well-commented to explain the rationale between feature engineering choices, model selection, data preprocessing and hyperparameter tuning. The decisions of the Fugacity 2026 organizing committee and the judges will be final and binding. There’s zero tolerance to plagiarism. All code must be original.  Unless explicitly allowed in the problem statement, the use of external proprietary datasets or unverified pre-trained weights is strictly prohibited. Only standard open-source frameworks like Scikit-Learn, PyTorch, TensorFlow etc are permitted. Shortlisted teams must confirm their online presence within 48 hours of result declaration. Arrangements for an online presentation can be made only under certain circumstances on a case-by-case basis. Important Disclaimer Travel & Lodging: Please note that no accommodation or transportation arrangements will be provided by the Fugacity 2026 team. Attendees coming from outside Kharagpur must organize and fund their own transit and lodging. Prizes and Recognition Winners and Runners-up will receive Merit Certificates + Cash Prizes. All finalists will receive finalist certificates.