Christian Weilbach
Founder and CEO, Simmis
christian@weilbach.name christian.weilbach.name github.com/whilo Google Scholar
Education
- 2018–2025 PhD in Computer Science, University of British Columbia. Structured amortized variational inference, with Frank Wood.
- 2015–2018 Master's in Computer Science, Heidelberg University. Dictionary learning with Bayesian GANs for few-shot classification.
- 2011–2015 Bachelor's in Computer Science, Heidelberg University. An online learning algorithm for LIF-based Boltzmann machines.
- 2006–2011 Philosophy and cultural anthropology, Heidelberg University.
- 2005–2006 Mathematics and computer science, Heidelberg University.
Grants, honours and awards
- 2024 Oral presentation at ICML: All-in-one simulation-based inference.
- 2023 Oral presentation at ICML (top 2.4%): Graphically structured diffusion models.
- 2023 Google research grant of over 80,000 USD for graphically structured diffusion models, with Frank Wood.
- 2016 Best poster award, EuroSys, London (replikativ: open P2P replication with CRDTs).
Publications
- 2025 Y. He, C. Weilbach, M. E. Wojciechowska, Y. Zhang, F. Wood. PLAICraft: Large-Scale Time-Aligned Vision-Speech-Action Dataset for Embodied AI. arXiv 2025.
- 2025 C. Weilbach, F. Wood. Daphne: Multi-Pass Compilation of Probabilistic Programs into Graphical Models and Neural Networks. TMLR 2025.
- 2024 M. Gloeckler, M. Deistler, C. Weilbach, F. Wood, J. H. Macke. All-in-one Simulation-Based Inference. ICML 2024. Oral.
- 2024 R. Fayyazi, C. Weilbach, F. Wood. Prospective Messaging: Learning in Networks with Communication Delays. arXiv 2024.
- 2023 C. Weilbach, W. Harvey, F. Wood. Graphically Structured Diffusion Models. ICML 2023. Oral, top 2.4%.
- 2023 A. Campbell, W. Harvey, C. Weilbach, V. De Bortoli, T. Rainforth, A. Doucet. Trans-Dimensional Generative Modeling via Jump Diffusion Models. NeurIPS 2023.
- 2023 C. Weilbach, W. Harvey, H. Shirzad, F. Wood. Scaling Graphically Structured Diffusion Models. ICML workshop on structured probabilistic inference and generative modeling 2023.
- 2023 G. Gonzalez Garcia, C. Weilbach. If the Sources Could Talk: Evaluating Large Language Models for Research Assistance in History. Computational Humanities Research 2023.
- 2022 W. Harvey, S. Naderiparizi, V. Masrani, C. Weilbach, F. Wood. Flexible Diffusion Modeling of Long Videos. NeurIPS 2022.
- 2022 F. Wood, A. Warrington, S. Naderiparizi, C. Weilbach, V. Masrani, W. Harvey, et al.. Planning as Inference in Epidemiological Dynamics Models. Frontiers in Artificial Intelligence 2022.
- 2021 J. Weilbach, S. Gerwinn, C. Weilbach, M. Kandemir. Inferring the Structure of Ordinary Differential Equations. ICML time series workshop 2021.
- 2021 B. Beronov, C. Weilbach, F. Wood, T. Campbell. Sequential Core-Set Monte Carlo. UAI 2021.
- 2020 C. Weilbach, B. Beronov, F. Wood, W. Harvey. Structured Conditional Continuous Normalizing Flows for Efficient Amortized Inference in Graphical Models. AISTATS 2020.
- 2019 C. Weilbach, B. Beronov, W. Harvey, F. Wood. Efficient inference amortization in graphical models using structured continuous conditional normalizing flows. Symposium on Advances in Approximate Bayesian Inference.
- 2017 C. Weilbach, A. Bieniusa. Time-sensitive probabilistic inference for the edge. technical report, arXiv 2017.
- 2016 C. Weilbach, K. Kühne, A. Bieniusa. Decoupling Conflict Resolution with CDVCS. PaPoC workshop at EuroSys 2016. With the replikativ poster, best poster award at EuroSys 2016.
- 2016 C. Weilbach, K. Kühne, A. Bieniusa. replikativ: open P2P replication with CRDTs. Poster, EuroSys, London.
Talks
- 2026 Rationalizing a Society of Learners. Max Planck Institute for Intelligent Systems, Tübingen.
- 2026 Programming as and for inference. London Clojurians.
- 2025 Partially evaluating higher-order probabilistic programs without stochastic recursion to graphical models. LAFI workshop at POPL.
- 2024 Flexible diffusion models for videos, molecules, and inference in structured simulators. Google DeepMind, London.
- 2023 Video and graphically structured diffusion models. OpenAI, San Francisco.
- 2023 Graphically structured diffusion models. Mila, Montréal.
- 2019 Anglican: Turing-complete functional probabilistic programming as a domain-specific language. Trends in Functional Programming.
- 2019 Datalog as a distributed systems language. Clojure Meetup Vancouver.
- 2017 Let's replicate state together. clojureD, Berlin.
- 2017 Machine learning workshop. CubaConf, Havana.
Projects
- 2025– Simmis: a workspace in which people and AI agents share one memory (chat, wiki, knowledge bases, double-entry accounts, code), with work done in branches and adopted by review.
- 2026 Klara: a language for recursive self-improvement. Running programs keep versioned worlds, specialise themselves with staged, untrusted optimisers, and admit changes through a small kernel with proofs checked in Ansatz.
- 2026 Ansatz, a dependently typed Clojure DSL with a Lean 4 compatible kernel; Wandler, verified transducer optimisation.
- 2026 Foerster: probabilistic inference in forkable worlds (SMC, particle MCMC, HMC, variational inference, counterfactuals), and a data-grounded simulation of Stuttgart with 936,597 agents and Bayesian inference over its parameters.
- 2026 Raster: functional numerical computing that compiles one algebra to each device, with ports of UMAP, EVoC and pretrained Hugging Face models to the JVM.
- 2026 Spindel, a reactive runtime with forkable state; partial-cps; Dvergr, an agent framework with forkable memory and measured workflows.
- 2026 Datahike ecosystem: Yggdrasil, Proximum, Stratum, Scriptum, Geschichte and pg-datahike.
- 2026 Kontor, an accounting kernel; Superficie, a readable surface syntax for Clojure; contributions to PyMC-Marketing.
- 2026 Open migration databases with Dr. Giselle González García: the Cuba–Ireland Immigration Database and Redes Isleñas, built on Datahike and Spindel.
- 2023– PLAICraft: large-scale multimodal data collection of multi-agent Minecraft play. Led 11 volunteers and several students on architecture and code.
- 2019–2022 Daphne: a compiler from first-order probabilistic programs to graphical models and structured normalizing flows; adapted for teaching a graduate course in probabilistic programming at UBC.
- 2018 Datahike: design and development of a Datalog database on persistent data structures.
- 2017 A reverse-mode autograd for Clojure.
- 2016 Bayesian dropout for recursive neural networks in sentiment analysis; dropout as a Gaussian process approximation; complex knowledge-base embeddings.
- 2015 Research assistant in the Human Brain Project, extending my Bachelor thesis; a Boltzmann machine library for contrastive divergence with STDP.
- 2013– replikativ: distributed state management with CRDTs.
Commercial work
- 2025– Founder and CEO of Simmis.
- 2025 Python port and modernisation of the PCA implementation of pol.is.
- 2019– Consulting for the deployment of Datahike at the Swedish Public Employment Service.
- 2018– Founder of LambdaForge UG.
- 2018 Improvements to the probabilistic programming system Anglican.
- 2016–2017 Design and implementation of a backend and apps for feebame UG.
- 2016 Reachability analysis and callee prediction with Bayesian deep learning for Keatec GmbH, Frankfurt.
- 2014–2015 Statistical analysis of medical data for BMI Heidelberg.
Teaching and community
- 2020 Organiser, Vancouver Julia Meetup.
- 2017–2018 Teaching assistant, Master's lecture “Fundamentals of Machine Learning”, Heidelberg University.
- 2017 Cornell, Maryland, Max Planck Pre-doctoral Research School, Saarbrücken.
- 2015–2018 Co-organiser, Clojure Meetup Mannheim-Heidelberg; organiser, Machine Learning Meetup Rhein-Neckar.
Languages
German (native), English (fluent), French (basic).