Session Host
Tiffa Foster

Tiffa is a playwright and technologist who enjoys bridging narrative semantics and machine learning. She studied Bioethics, Philosophy, Mandarin, and Tibetan at UVA, later completing Harvard Medical School’s Media & Medicine program. While her writing, spanning an opera, TV pilots, and a publication in The Best Women’s Stage Monologues 2024, has been produced globally, an AI co-authored novel project pulled her back into code. She loves writing Kaggle Benchmark tasks, building Python stress-tests against LLMs, and presenting AI safety white papers. She is fascinated by tokenizer fragility, and her favorite Unicode character is U+A66E.
Sessions at this workshop:
| Thurs, 14:40 | The Tokenizer Comedy: A Catalog of Joyful Failures in LLM Unicode Processing |