Observed arrival · 2026-11-08
Daniel Tsen’s walkable research city
A research portfolio presented as an interactive city, with buildings leading to work in urban simulation, peptide machine learning, astronomy, and engineering.
- For
- Research readers, urban-simulation builders, and prospective collaborators
- Worth noticing
- The peptide paper notes oversampling occurred before the split, so its reported 0.996 cross-validated AUC is treated as an upper bound.
Field notes
The urban simulation includes 500 LLM-driven pedestrians in Taipei’s Da’an District, with a social gate that can prompt agents to stop and deliver canned facts. Tsen says the browser model is a toy, not lab data, and describes deterministic replay at 10-minute intervals. Elsewhere, the peptide section reports an AUC of 0.996 while disclosing that oversampling preceded the train/test split and treating the result as an upper bound.
Observed signals
Read the marks
Editorial observations of this landing page, not a rating.
One card from the complete issue