Core Technology
Foundational methods for reasoning, probabilistic inference, and model-building in complex environments.
About
I’m a postdoctoral research scientist at Basis, a non-profit research institute, working with Matt Levine (Basis) and Youssef Marzouk (MIT). I’m largely interested in Bayesian machine learning and statistics, causal inference, and dynamical systems, and especially like the many intersections of these topics.
I previously completed my PhD at Stony Brook University working with Petar Djurić, where my dissertation focused on online and sequential Bayesian learning, with applications in ensembles and experimental design. During Summer 2025, I was a research intern at Basis, where I worked on problems in applied Bayesian ML and causality. During my PhD, I was a member of the Southwest Integrated Field Laboratory, where I worked with Katia Lamer in applying ML techniques to applied experimental design problems.
Projects
Current and recent Basis projects.
Cities
Participatory city models that help residents, community groups, and policymakers reason about uncertain causes and policy consequences.
Software
Tools for dynamical systems and scientific machine learning.
Dynestyx
Dynestyx is an extension of NumPyro for first-class support of dynamical systems, with a unified interface to structured inference methods for state-space models.
Recent Publications
Recent papers and preprints.
Sequential Inference for Gaussian Processes: A Signal Processing Perspective
D. Waxman, Fernando Llorente, Petar M. Djurić
Designing an Optimal Sensor Network via Minimizing Information Loss
D. Waxman, Fernando Llorente, Katia Lamer, Petar M. Djurić
Bayesian Ensembling: Insights from Online Optimization and Empirical Bayes
D. Waxman, Fernando Llorente, Petar M. Djurić
Tangent space causal inference: Leveraging vector fields for causal discovery in dynamical systems
Kurt Butler, D. Waxman, Petar M. Djurić
Articles
Basis essays and updates this person wrote or contributed to.
Building an Unverified Compiler with Agents
Four agents spent 14 days and 93,000 lines of Lean building a verified JS-to-WASM compiler from scratch. The compiler ran; the proofs didn’t close.