ML+Coffee: How Can I Apply ML/AI To My Data?

Please fill out the registration form if you plan to attend (light refreshments provided).

Summary: The monthly ML+Coffee social brings together ML practitioners across campus so that we can connect with one another, discuss and work on ML projects, and enjoy some light (caffeinated) refreshments ☕. Attendees are encouraged to bring their laptops and/or any questions about ML. Feel free to show up late and/or leave early if you can't stay for the full two hours.

• 9:00-9:30am (Networking): Grab a coffee and meet fellow ML/AI practitioners on campus.
• 9:30-10:00am (Discussion / Seeking Help): Chunxiao Jing — PhD Student. Hello everyone! I'm in the Department of Agricultural and Applied Economics, focusing on economic development in Africa. Our team is working on a climate–conflict project using pixel-level panel data across the continent. A core challenge is forecasting where conflict will occur and, equally important, understanding which climate variables (e.g., rainfall shocks, temperature extremes, drought indices) most strongly drive those risks. We'd like advice on applying machine learning to this setting—especially methods that handle spatiotemporal structure, class imbalance (rare conflict events), and out-of-sample validation over time. Our goals are to (1) build models that predict conflict from climate features one year ahead at the grid-cell level, and (2) interpret feature effects and directions to clarify the relationships between climate shocks and conflict incidence. If you have recommendations on model choices, calibration/thresholding, feature engineering, or best practices for explainability with pixel-level data, we'd love to hear them at ML+Coffee!
• 10:00-11:00am (Demo/Presentation): Algorithmic Prompt Design and Application — Dhruba Jyoti (DJ) Paul, ML Engineer @ All3D. Learn about advanced prompt engineering strategies, prompt inversion, and other methods that can help you build better LLM-based applications (including RAG).