About · Research
From a model's hidden uncertainty to human oversight.
I am a dedicated AI researcher passionate about building trustworthy AI systems that can make the future better. I am currently pursuing my PhD (2023–2027, expected) under the joint supervision of Prof. Elliott Ash and Prof. Mrinmaya Sachan at ETH Zurich, and Prof. Markus Leippold at University of Zurich. I divide my time equally between both institutions.
I build AI systems that know when to be trusted: I extract the uncertainty a model already carries and expose it for human supervision. One requirement sits behind everything — a model should report low confidence whenever it is likely to be wrong — and that single reporting ability is what makes oversight possible. What changes is how tractable the problem is, and that turns on whether the uncertainty is objectively checkable.
These methods matter most where the stakes are high and people stay accountable for the answer — which is why I ground them in climate and law. From corporate climate disclosures to legal reasoning, an unverified claim can sway policy or judgment, so domain experts need to see exactly how far to trust a model. I like to aim high while keeping my feet firmly on the ground.
Verifiers that report confidence
Verifying a math reasoning step, or whether retrieved evidence is relevant: labels are cheap, models are strong, and uncertainty can be trained and rigorously tested. Here I build non-agentic verifiers like ReProbe, which reads a model's internal states mid-reasoning to tell when an answer has settled.
Surfacing contested judgments
Subjective classification and conflicting human labels — signal, not noise. I build uncertainty-aware classifiers, evidence-grounded specialists whose every claim traces to its source, and interfaces that surface the edge cases people must adjudicate.
Projects & community service
Open models, climate NLP, and research communities.
Community collaboration
Apertus: Democratizing Open and Compliant LLMs
Community collaboration on democratizing open and compliant LLMs for global language environments, with contributions to trustworthiness post-training.
Technical report →Community collaboration
When AI Benchmarks Plateau
Community collaboration accepted to ICML 2026 on benchmark saturation and how plateauing scores affect evaluation practice.
ICML 2026 →Workshop
ClimateNLP at ACL 2025
Organizing the second ClimateNLP workshop at ACL 2025, Vienna.
Workshop
ClimateNLP at ACL 2024
Organized the first ClimateNLP workshop at ACL 2024, Bangkok.
Publications
Selected research
Mining uncertainty from language models — and the verifiers, annotations, and interfaces that surface it for people.
ReProbe: Efficient Test-Time Scaling of Multi-Step Reasoning by Probing Internal States of Large Language Models
Reads the model's own internal states mid-reasoning to tell when an answer has settled, scaling test-time reasoning only as far as the model actually needs.
Can Reasoning Help Large Language Models Capture Human Annotator Disagreement?
Tests whether reasoning lets a model recognize when a question is genuinely contested, instead of collapsing real human disagreement into one overconfident answer.
DIRAS: Efficient LLM Annotation of Document Relevance for Retrieval Augmented Generation
Trains small, cheap models to judge whether retrieved documents are truly relevant, giving RAG an external check on its evidence before the generator ever sees it.
Co-DETECT: Collaborative Discovery of Edge Cases in Text Classification
Pairs human annotators with an LLM to surface the edge cases a classifier quietly gets wrong, pointing oversight straight at the model's blind spots.
Towards Faithful and Robust LLM Specialists for Evidence-Based Question-Answering
Trains open QA specialists to answer strictly from the evidence they are given and resist being misled when it is noisy or missing — keeping every answer traceable to its source.
AFaCTA: Assisting the Annotation of Factual Claim Detection with Reliable LLM Annotators
Cross-checks several LLM annotators against each other to reliably flag which sentences are checkable factual claims — the first filter any fact-verification pipeline needs.
Supervised Master Thesis & Mentorship
Unlocking LLM Legal Reasoning with IRAC-Constrained Chain-of-Thought
Master student · Adam Rahmoun
Full publication list
Accepted In review / preprint
Education
Academic path
PhD @ ETH D-GESS
ETH Zürich & University of Zürich
MSc in Data Science & Machine Learning
University College London
BEng in Computer Science
University of Hong Kong