Robin Chan
Robin Chan is a PhD student at the Institute for Machine Learning at ETH Zürich, advised by Prof. Ryan Cotterell at Rycolab. His research focuses on inference-time language-model control, integrating methods from probabilistic inference and human-computer interaction.
He is active in the GenLM consortium, building an open-source ecosystem for language-model probabilistic programming. Before his PhD he obtained an M.Sc in Data Science at ETH Zürich, with a thesis on controlled LLM code generation at IBM Research Europe.
Robin won Best Paper at CHI 2026 (PleaSQLarify, top 1% of ~7,000 submissions) and an Outstanding Reviewer Award at NeurIPS 2025.
Affiliation: Institute for Machine Learning, ETH Zürich (Rycolab).
- CHI 2026 Best Paper Award (top 1%)
- NeurIPS 2025 Outstanding Reviewer
Robin's prior work.
These papers are notsubmissions to the Mental Models of AI workshop. The workshop hasn't happened yet. They are the body of work Robin brings to the committee, verified against personal sites and Google Scholar.
PleaSQLarify: Visual Pragmatic Repair for Natural Language Database Interfaces
Reframes ambiguity in natural-language database interfaces as a pragmatics problem and introduces pragmatic repair, incremental clarification through minimal interaction, implemented through a visual UI of interpretable decision variables. A study with twelve participants shows users recognise alternative interpretations and resolve ambiguity efficiently.
Ensembling Language Models with Sequential Monte Carlo
A unified framework for composing K language models into f-ensemble distributions, sampled with a byte-level sequential Monte Carlo algorithm operating in a shared character space, enabling consistent ensembling across models with mismatching vocabularies.
