
Aug 2026
Instructing at the AI in Practice Summer Institute
Five days, 9 to 5, in a room of researchers, postdocs, PhD students, and faculty working through what it actually takes to build and evaluate AI models on their own problems. The sequence track was mine to lead: RNNs and LSTMs on time series, then transformers and the attention mechanism, built up from why a fixed context vector runs out of room, through query, key, and value, to the full scaled dot-product. Thanks to Valentina Staneva, Anthony Arendt, and the eScience team for putting it together.







