AI helped you train your ML model, but can you vouch for it?

🥁 Hackathon Challenge AI can write a complete machine learning pipeline in seconds. But can you vouch for it? In this half-day hackathon, co-hosted by Probabl and the GLA Data for London Team as part of London Data Week 2026, participants will tackle real policy problems faced by the Greater London Authority, including youth employment, building decarbonisation, and open data discovery. The focus isn't just on training a model, but on being able to evaluate, explain, and vouch for it – a skill that's only becoming more important as AIs can generate ML pipelines in seconds. Participants will be encouraged to use the Python library skore in their evaluations. The winning teams will get to pitch their solution to the GLA’s data leadership team in the afternoon. Note: Participants are asked to bring their own laptop. 🗓️ Agenda 9:00-9:30: Doors open Doors open at 9:00. We ask all participants to arrive before 9:30 so that we can start the fun on time. 9:30-10:00: Welcome We’ll start with welcome speeches and a demo of skore. Participants will be asked to form teams of 2-4 and select the policy problem they want to tackle during the hackathon. The options include: Can we help London’s young people find work, education, or training? How do we decarbonise London’s buildings? How can we improve searching for London’s data? 10:00-12:00: Hackathon Teams tackle their chosen policy problem using provided datasets. Teams will be assessed on how well they can explain and evaluate their model and decision-making – whether the code was hand-written, AI-generated, or both. 12:00-13:00: Presentations & prizes. Each team gives a short presentation of their approach. Prizes are given to the winning team per policy problem. 🏆 Prizes and Perks • Winning teams (1x per policy problem): 100% discount code for Skolar certification + pitch solution to the GLA's data leadership team in the afternoon • All participants: 100% discount code for the Skolar Discovery certification - a tangible credential for the CV and LinkedIn account, regardless of their hackathon result. • Refreshments and lunch provided • Networking with fellow data science practitioners and the Probabl and GLA data teams