Machine Learning Marathon
The Machine Learning Marathon (MLM) is an annual 12-week ML/AI hackathon hosted by the ML+X community at UW-Madison. Each fall, teams of 2-5 tackle real-world challenges sourced from UW researchers, industry partners, and the broader Kaggle community. Challenges span a range of domains and difficulty levels – from beginner-friendly to competitive challenges with prizes.
Below you’ll find challenges from the current Marathon alongside editions from past years – use the category filters (e.g., MLM26, MLM25) to browse by season. Most challenges remain open on Kaggle even after their Marathon season ends, making them great self-study material any time of year. Dive into the datasets, explore the discussion tabs for tips and solution writeups from past participants, and benchmark your own approaches at your own pace.
Want to participate live? Join ML+X and watch for the next Marathon registration each September. Interested in proposing a challenge? Reach out to Chris Endemann.
Explore challenges, past and present
Badger Scribe: Transcribing 19th-Century Handwritten Archives on a Single GPU
Badger Code: Building a Local Coding Agent on a Single GPU
WattBot 2026: Investigating AI’s Environmental Footprint with RAG
Snapshot Wisconsin: Oh Deer – Predicting Population Dynamics from Trail Cameras
Fixing the Poisoned Well: Reverse-Engineering a Backdoored Streamflow Model
Clustering the BioTrove Dataset
Brain-to-Text ’25: Decoding Speech from Neural Activity
MaveDB: Protein Variant Effect Prediction
WattBot 2025: Estimating AI Emissions with RAG
Natural Language Processing with Disaster Tweets
UW-Madison GI Tract Image Segmentation
