Dennis Frauen
I am a Research Scientist at Meta on the Central Applied Science (CAS) team. Previously I completed my PhD at LMU Munich in the ELLIS program, co-supervised by Stefan Feuerriegel (LMU Munich) and Mihaela van der Schaar (University of Cambridge), where I also co-directed the CausalML Lab.
I am interested in methodology for building and evaluating AI models — methods that are of high practical importance while being principled in providing guarantees, often grounded in statistics and causal inference.
My work is organized around three key themes: (I) Efficiency, (II) Robustness, and (III) Trustworthiness. Specific research areas include:
- LLM and agent evaluation and post-training
- (Causal) tabular foundation models
- Off-policy learning and reinforcement learning
- Causal inference and debiased machine learning
- Uncertainty quantification (e.g., conformal prediction, probabilistic and Bayesian methods)
News!
[24/09/2026] Three papers accepted at NeurIPS 2026! Assessing the robustness of heterogeneous treatment effects in survival analysis under informative censoring, Generalized Bayes for Causal Inference, and Targeted Synthetic Control Method.
[04/05/2026] I started a new position as a Research Scientist at Meta and moved to NYC!
[30/04/2026] Three papers accepted at ICML 2026! Nonparametric LLM Evaluation from Preference Data, Rank-Learner: Orthogonal Ranking of Treatment Effects, and DeepBlip: Estimating Conditional Average Treatment Effects Over Time.
[24/03/2026] I succesfully passed my PhD defense with distinction (summa cum laude)!
[27/01/2026] Six papers accepted at ICLR 2026! Foundation Models for Causal Inference via Prior-Data Fitted Networks, Overlap-weighted orthogonal meta-learner for treatment effect estimation over time, An Orthogonal Learner for Individualized Outcomes in Markov Decision Processes, G-Transformer for Conditional Average Potential Outcome Estimation over Time, Efficient and Sharp Off-Policy Learning under Unobserved Confounding, and Overlap-Adaptive Regularization for Conditional Average Treatment Effect Estimation.
[10/01/2026] Our paper Orthogonal Representation Learning for Estimating Causal Quantities got accepted as an oral presentation at the AISTATS 2026!
[09/10/2025] Our paper How reliable are treatment effects in clinical trials with dropout? got accepted as an oral presentation at the CauScien Workshop of NeurIPS 2025!
[19/09/2025] Five papers accepted at NeurIPS 2025! Orthogonal Survival Learners for Estimating Heterogeneous Treatment Effects from Time-to-Event Data, LLM-Driven Treatment Effect Estimation Under Inference Time Text Confounding, Treatment Effect Estimation for Optimal Decision-Making, Conformal prediction for causal effects of continuous treatments, and Improving the Generation and Evaluation of Synthetic Data for Downstream Medical Causal Inference.
[16/06/2025] I started my research internship at Meta and moved to NYC!
[27/05/2025] Our article Algorithms for reliable decision-making need causal reasoning got accepted at Nature Computational Science!
[06/05/2025] Our paper Learning Representations of Instruments for Partial Identification of Treatment Effects got accepted at ICML 2025!
[08/04/2025] I was named a Rising Star in AI by King Abdullah University of Science and Technology (KAUST) and presented our research in Thuwal, Saudi Arabia.
[23/01/2025] I am excited to join the van der Schaar lab at the University of Cambridge as a visiting PhD student for the next three weeks as part of the ELLIS PhD program. If you are in the Cambridge/ London area and would like to meet for a coffee, let me know!
[22/01/2025] Two accepted papers at ICLR 2025! Model-Agnostic Meta-Learners for Estimating Heterogeneous Treatment Effects over Time and Constructing Confidence Intervals for Average Treatment Effects from Multiple Datasets.
[20/11/2024] I won the G-Research Early Career Grant. It will support my research stay with Mihaela van der Schaar at the University of Cambridge in January/February 2025.
[30/10/2024] We launched the CausalML lab at LMU Munich, where I will be co-director. The project is supported by the German Federal Ministry of Education and Research, in collaboration with Niki Kilbertus (Helmholtz/TUM), Stefan Bauer (Helmholtz/TUM), Nadja Klein (KIT), and EconomicAI.
[03/06/2024] I moved to the San Francisco Bay area to start my PhD internship at Netflix Research. I will be working on deep learning for message recommendations, leveraging tools from Causal ML and Reinforcement learning.
[16/05/2024] Our paper Causal Machine Learning for Cost-Effective Allocation of Development Aid was accepted at KDD 2024.
[02/05/2024] Two accepted papers at ICML 2024! Fair Off-Policy Learning from Observational Data and Meta-Learners for Partially-Identified Treatment Effects Across Multiple Environments (preprint coming soon).
[19/04/2024] Our perspective paper Causal machine learning for predicting treatment outcomes was published at Nature Medicine.
[16/01/2024] Four accepted papers at ICLR 2024! A Neural Framework for Generalized Causal Sensitivity Analysis, Bayesian Neural Controlled Differential Equations for Treatment Effect Estimation, Causal Fairness under Unobserved Confounding: A Neural Sensitivity Framework, and Bounds on Representation-Induced Confounding Bias for Treatment Effect Estimation.
[01/11/2023] I received a mention as a top reviewer at NeurIPS 2023.
[26/09/2023] Together with Stefan and Abdu, we presented an overview of our research at the Ladenburger Diskurs.
[21/09/2023] Three accepted papers at NeurIPS 2023! Sharp Bounds for Generalized Causal Sensitivity Analysis, Partial Counterfactual Identification of Continuous Outcomes with a Curvature Sensitivity Model, and Reliable Off-Policy Learning for Dosage Combinations.
[24/08/2023] I presented some of our ongoing work on Causal sensitivity analysis at Microsoft Research, Cambridge.
[11/06/2023] I am excited to join the van der Schaar lab at the University of Cambridge as a visiting PhD student this summer.
[24/04/2023] Our paper Normalizing Flows for Interventional Density Estimation was accepted at ICML 2023.
[22/01/2023] Our paper Estimating individual treatment effects under unobserved confounding using binary instruments was accepted at ICLR 2023.
[19/11/2022] Our paper Estimating average causal effects from patient trajectories was accepted at AAAI 2023.
[8/11/2022] I presented some of our ongoing work on Fair policy learning from observational data at the Causal Data Science Meeting 2022.