May 15, 2026 · people
Katelyn Crumpacker defends her M.Sc. thesis
Congratulations to Katelyn Crumpacker on defending her M.Sc. thesis, "LLM-Guided Runtime Parameter Optimization for Energy-Efficient Model Inference." The runtime knobs that decide what inference costs in energy are numerous, interacting, and rarely tuned. Her work uses a language model to search that space.