Research Directions

Our group works on machine learning for science, grounded in methodology for uncertainty quantification, sampling, optimization, and active learning. We are particularly excited to apply these methods to domains such as chemistry, biology, and physics. Our group has priority access to a large cluster of NVIDIA B200/B300 GPUs, supporting large-scale training and simulation.

Topics include:

  • variational inference [1, 2, 3, 4]
  • generative modeling [6]
  • sampling [5, 7]
  • optimization for machine learning [8]
  • active learning and experimental design [9, 10]
  • simulation-based inference [11]
  • model misspecification [12, 13]
  • chemistry and materials applications [9]
  • biology applications [10]
References (click to show)(click to hide) (* co-lead author, ^ co-senior author)
  1. D. Cai, C. Modi, L. Pillaud-Vivien, C. C. Margossian, R. M. Gower, D. M. Blei, L. K. Saul. Batch and match: black-box variational inference using a score-based divergence. ICML, 2024.
  2. C. Modi*, D. Cai*, L. K. Saul. Batch, match, and patch: low-rank approximations for score-based variational inference. AISTATS, 2025.
  3. D. Cai, C. Modi, C. C. Margossian, R. M. Gower, D. M. Blei, L. K. Saul. EigenVI: score-based variational inference with orthogonal function expansions. NeurIPS, 2024.
  4. D. Cai, R. M. Gower, D. M. Blei, L. K. Saul. Fisher meets Feynman: score-based variational inference with a product of experts. NeurIPS, 2025.
  5. D. M. Zoltowski, D. Cai, R. P. Adams. Slice sampling reparameterization gradients. NeurIPS, 2021.
  6. L. Appleby, D. Cai, L. K. Saul, D. D. Lee. Div, grad, curl, and all that: electrostatic models for score matching. Preprint, 2026.
  7. D. Cai, R. P. Adams. Multi-fidelity Monte Carlo: a pseudo-marginal approach. NeurIPS, 2022.
  8. J. Ko, T. Parshakova, D. Cai^, R. M. Gower^. SoftServe: a practical quasi-Newton method for deep learning. Preprint, 2026.
  9. A. Novick*, D. Cai*, Q. Nguyen, R. Garnett, R. P. Adams, E. Toberer. Probabilistic prediction of material stability: integrating convex hulls into Bayesian active learning. Materials Horizons, 2024.
  10. A. Jones*, D. Cai*, D. Li, B. E. Engelhardt. Optimizing the design of spatial genomics studies. Nature Communications, 2024.
  11. X. Shen, D. Cai, C. Zhang, D. M. Blei. Simulation-based empirical Bayes. Preprint, 2026.
  12. D. Cai*, T. Campbell*, T. Broderick. Finite mixture models do not reliably learn the number of components. ICML, 2021.
  13. D. Cai, T. Campbell, T. Broderick. Edge-exchangeable graphs and sparsity. NeurIPS, 2016.

Join the Lab

We’re growing! I’m excited to recruit PhD students and postdocs who are interested in AI for science and probabilistic machine learning.

Below is the current status of openings. I’ll update this page as new positions become available.

PhD students: Please apply to the Cornell Computer Science PhD program and list my name.

Postdocs: Please check back for available positions soon.

Undergrads and masters students: At this time, there are no available positions.

Interns, remote students, etc.: I do not have any available positions.

Due to the high volume of emails I receive, I won’t be able to respond to individual inquiries.