Reasoning and decision making under uncertainty
I am a postdoctoral researcher at the Telecooperation Lab, TU Darmstadt, and an affiliated postdoctoral researcher at Tongji University, Shanghai. I completed my PhD in Computer Science at TU Darmstadt in 2025 (summa cum laude), supervised by Prof. Max Mühlhäuser.
My research focuses on reliable reasoning and decision making under uncertainty, particularly conformal prediction, calibration transfer, and reliability under distribution and model shift. My broader work includes neuro-symbolic learning, LLM reasoning, and model evaluation.
I am currently studying how uncertainty can guide human–AI decision making. Emerging directions include possibilistic and imprecise representations of epistemic uncertainty, and game-theoretic models of repeated human–AI interaction.
News
- Our project 4d.City: 4-Dimensional Counterfactual Urban World Models won the Hessian AI Innovation Award.
- Received the Ingenium Postdoctoral Support Package (€3,000) from TU Darmstadt for Uncertainty Quantification for Reliable Human–AI Interaction, an independent project running from October 2026 to January 2027.
- Sep 2026 Auditing Conformal Prediction for Language Models Under Inference-Time Model Shift accepted at the NeurIPS 2026 Evaluations & Datasets Track.
- Sep 2026 Received the NeurIPS 2026 Top Reviewer Award.
- May 2026 Received the ICML 2026 Gold Reviewer Award.
- May 2026 Conformal Calibration Transfer accepted at ICML 2026.
- May 2026 Position: Uncertainty is a Strategic Signal in Human–AI Decision Making accepted at ICML 2026.
- Jul 2025 Began a research visit at Tongji University, Shanghai, focusing on LLM-based structured reasoning and decision making for urban planning.
- Jun 2025 PhD in Computer Science awarded summa cum laude at TU Darmstadt.
Selected Publications
-
NeurIPS 2026Auditing Conformal Prediction for Language Models Under Inference-Time Model Shift
Audits how changes in inference configuration between calibration and deployment affect conformal coverage, even when the trained model and task distribution remain unchanged.
-
ICML 2026Conformal Calibration Transfer
Transfers conformal calibration across input spaces under distribution shift, with finite-sample target-domain coverage guarantees and a diagnostic of transfer mismatch.
-
ICML 2026Position: Uncertainty is a Strategic Signal in Human–AI Decision Making
Argues for treating uncertainty as a strategic signal that shapes human–AI decisions, incentives, and adaptation over repeated interaction.
-
NAACL 2025CLEAR-Command: Coordinated Listening, Extraction, and Allocation for Emergency Response with Large Language Models
-
ICSE 2025MARQ: Engineering Mission-Critical AI-Based Software with Automated Result Quality Adaptation
-
CoRL 2024Conformal Prediction for Semantically-Aware Autonomous Perception in Urban Environments
-
IROS 2024NeSyMoF: A Neuro-Symbolic Model for Motion Forecasting
-
arXiv 2025SafePath: Conformal Prediction for Safe LLM-Based Autonomous Navigation
Research Projects & Funding
-
2026–2027
Uncertainty Quantification for Reliable Human–AI Interaction
Investigating uncertainty-aware policies for recommendation, verification, deferral, and information acquisition. The project supports a controlled human-participant study of decision quality, appropriate reliance, and human–AI team performance.
-
2022–2025
SemUrban — Semantically Enriched Deep Learning for Autonomous Systems
Led research on integrating structured knowledge into deep learning for autonomous systems, with responsibility for the scientific agenda, technical milestones, budget, stakeholder coordination, and research transfer.
Academic Service & Teaching
- Recognition: NeurIPS 2026 Top Reviewer; ICML 2026 Gold Reviewer Award.
- Reviewing: TMLR, ICML, NeurIPS, KDD, ICCV, NAACL, ICRA, CHI, and IEEE Robotics and Automation Letters.
- Supervision: Supervised more than 10 Bachelor’s and Master’s theses on uncertainty quantification, conformal prediction, neuro-symbolic learning, autonomous systems, LLM-based systems, and human–AI interaction.
- Teaching: Contributions to Distributed Algorithms and Ubiquitous Computing / Machine Learning, including student mentoring and project supervision.