Saifuddin Syed

Assistant Professor · UBC Statistics

I am an Assistant Professor in the Department of Statistics at the University of British Columbia, and an inaugural member of the AI Methods for Scientific Impact (AIM-SI) cluster within CAIDA.

My research focuses on the design and analysis of annealing algorithms for scalable Bayesian inference and generative modelling. I collaborate with scientists across astronomy, chemistry, and biotechnology, and am a member of the Algorithms and Inference Working Group for the Next Generation Event Horizon Telescope (ngEHT).

Previously, I was a Florence Nightingale Bicentenary Fellow at Oxford's Department of Statistics, and completed a postdoc under Arnaud Doucet and a PhD under Alexandre Bouchard-Côté.

Google Scholar CV Email

Research

Annealing algorithms

Markov chain Monte Carlo

Scalable Bayesian inference

Neural samplers

Generative modelling

Scaling limits and optimal tuning

Probabilistic programming

AI for science

Prospective Students

I am actively recruiting MSc and PhD students with very strong mathematical and programming ability.

Admission is handled by the department, not by me individually, so the way to reach me is through the application itself. Apply to the UBC Statistics graduate program and say in your statement that you are interested in working with me and why. Be specific: naming the work of mine that drew you in, and what you would want to do next, tells me far more than a general statement of interest.

A separate email is not necessary and will not affect the outcome. I receive more enquiries than I can answer, so I am usually unable to reply to them individually.

Publications

Doctoral thesis

Non-Reversible Parallel Tempering on Optimized Paths

University of British Columbia, 2022 · supervised by Alexandre Bouchard-Côté

Marshall Prize
SSC Pierre Robillard Award
CAIMS Cecil Graham Dissertation Award
ISBA Savage Award, Theory and Methods (Honourable Mention)

Thesis

2026

[18]

Optimized Annealed Sequential Monte Carlo Samplers

Saifuddin Syed, Alexandre Bouchard-Côté, Kevin Chern, Arnaud Doucet

Journal of the Royal Statistical Society Series B, 2026

arXiv DOI

@article{syed2026optimized,
  title   = {Optimized Annealed Sequential {M}onte {C}arlo Samplers},
  author  = {Syed, Saifuddin and Bouchard-C\^ot\'e, Alexandre and Chern, Kevin and Doucet, Arnaud},
  journal = {Journal of the Royal Statistical Society Series B: Statistical Methodology},
  year    = {2026},
  doi     = {10.1093/jrsssb/qkag082},
  note    = {arXiv:2408.12057}
}

[17]

Conditional Diffusion Sampling

Francisco M. Castro-Macías, Pablo Morales-Álvarez, Saifuddin Syed, Daniel Hernández-Lobato, Rafael Molina, José Miguel Hernández-Lobato

International Conference on Machine Learning (ICML), 2026

arXiv Code

@inproceedings{castromacias2026conditional,
  title     = {Conditional Diffusion Sampling},
  author    = {Castro-Mac\'ias, Francisco M. and Morales-\'Alvarez, Pablo and Syed, Saifuddin and Hern\'andez-Lobato, Daniel and Molina, Rafael and Hern\'andez-Lobato, Jos\'e Miguel},
  booktitle = {Proceedings of the 43rd International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  volume    = {306},
  year      = {2026},
  publisher = {PMLR},
  note      = {arXiv:2605.04013}
}

[16]

CREPE: Controlling Diffusion with Replica Exchange

Jiajun He, Paul Jeha^*, Peter Potaptchik^*, Leo Zhang^*, José Miguel Hernández-Lobato, Yuanqi Du, Saifuddin Syed^*†, Francisco Vargas^*†

International Conference on Learning Representations (ICLR), 2026

arXiv Project

@inproceedings{he2026crepe,
  title     = {{CREPE}: Controlling Diffusion with Replica Exchange},
  author    = {He, Jiajun and Jeha, Paul and Potaptchik, Peter and Zhang, Leo and Hern\'andez-Lobato, Jos\'e Miguel and Du, Yuanqi and Syed, Saifuddin and Vargas, Francisco},
  booktitle = {International Conference on Learning Representations ({ICLR})},
  year      = {2026},
  note      = {arXiv:2509.23265}
}

[15]

Accelerated Parallel Tempering via Neural Transports

Leo Zhang^*, Peter Potaptchik^*, Jiajun He^*, Yuanqi Du, Arnaud Doucet, Francisco Vargas, Hai-Dang Dau, Saifuddin Syed

International Conference on Learning Representations (ICLR), 2026

arXiv

@inproceedings{zhang2026accelerated,
  title     = {Accelerated Parallel Tempering via Neural Transports},
  author    = {Zhang, Leo and Potaptchik, Peter and He, Jiajun and Du, Yuanqi and Doucet, Arnaud and Vargas, Francisco and Dau, Hai-Dang and Syed, Saifuddin},
  booktitle = {International Conference on Learning Representations ({ICLR})},
  year      = {2026},
  note      = {arXiv:2502.10328}
}

[14]

De Novo Design of Protein Switches with Diffusion-Based Ensemble Sampling

Ali Omidi, Jiajun He, Jennifer M. Bui, Jörg Gsponer, Saifuddin Syed

Under review · bioRxiv 2026.07.20.739027

bioRxiv

@article{omidi2026denovo,
  title   = {De Novo Design of Protein Switches with Diffusion-Based Ensemble Sampling},
  author  = {Omidi, Ali and He, Jiajun and Bui, Jennifer M. and Gsponer, J\"org and Syed, Saifuddin},
  journal = {bioRxiv},
  year    = {2026},
  doi     = {10.1101/2026.07.20.739027}
}

2025

[13]

Amortized Sampling with Transferable Normalizing Flows

Charlie B. Tan^*, Majdi Hassan^*, Leon Klein, Saifuddin Syed, Dominique Beaini, Michael M. Bronstein, Alexander Tong, Kirill Neklyudov

Advances in Neural Information Processing Systems (NeurIPS), 2025

arXiv

@inproceedings{tan2025amortized,
  title     = {Amortized Sampling with Transferable Normalizing Flows},
  author    = {Tan, Charlie B. and Hassan, Majdi and Klein, Leon and Syed, Saifuddin and Beaini, Dominique and Bronstein, Michael M. and Tong, Alexander and Neklyudov, Kirill},
  booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
  volume    = {38},
  pages     = {94290--94325},
  year      = {2025},
  note      = {arXiv:2508.18175}
}

[12]

Pigeons.jl: Distributed Sampling from Intractable Distributions

Nikola Surjanovic, Miguel Biron-Lattes, Paul Tiede, Saifuddin Syed, Trevor Campbell, Alexandre Bouchard-Côté

Proceedings of the JuliaCon Conference, 7:1–13, 2025

arXiv Code

@article{surjanovic2025pigeons,
  title   = {Pigeons.jl: Distributed Sampling from Intractable Distributions},
  author  = {Surjanovic, Nikola and Biron-Lattes, Miguel and Tiede, Paul and Syed, Saifuddin and Campbell, Trevor and Bouchard-C\^ot\'e, Alexandre},
  journal = {Proceedings of the JuliaCon Conferences},
  volume  = {7},
  pages   = {1--13},
  year    = {2025},
  doi     = {10.21105/jcon.00139},
  note    = {arXiv:2308.09769}
}

[11]

The Cosine Schedule is Fisher-Rao-Optimal for Masked Discrete Diffusion Models

Leo Zhang, Saifuddin Syed

Preprint · arXiv:2508.04884, 2025

arXiv

@misc{zhang2025cosine,
  title         = {The Cosine Schedule is {F}isher-{R}ao-Optimal for Masked Discrete Diffusion Models},
  author        = {Zhang, Leo and Syed, Saifuddin},
  year          = {2025},
  eprint        = {2508.04884},
  archivePrefix = {arXiv},
  primaryClass  = {stat.ML}
}

[10]

Reproducible Sampling from Intractable Distributions with Pigeons.jl

Nikola Surjanovic, Miguel Biron-Lattes, Paul Tiede, Saifuddin Syed, Trevor Campbell, Alexandre Bouchard-Côté

CODEML Workshop, International Conference on Machine Learning, 2025

Code

@inproceedings{surjanovic2025reproducible,
  title     = {Reproducible Sampling from Intractable Distributions with Pigeons.jl},
  author    = {Surjanovic, Nikola and Biron-Lattes, Miguel and Tiede, Paul and Syed, Saifuddin and Campbell, Trevor and Bouchard-C\^ot\'e, Alexandre},
  booktitle = {Championing Open-source Development in Machine Learning Workshop, International Conference on Machine Learning ({ICML})},
  year      = {2025}
}

[9]

Generalised Parallel Tempering: Flexible Replica Exchange via Flows and Diffusions

Leo Zhang, Peter Potaptchik, Arnaud Doucet, Hai-Dang Dau, Saifuddin Syed

Frontiers in Probabilistic Inference Workshop, International Conference on Learning Representations, 2025

arXiv

@inproceedings{zhang2025generalised,
  title     = {Generalised Parallel Tempering: Flexible Replica Exchange via Flows and Diffusions},
  author    = {Zhang, Leo and Potaptchik, Peter and Doucet, Arnaud and Dau, Hai-Dang and Syed, Saifuddin},
  booktitle = {Frontiers in Probabilistic Inference Workshop, International Conference on Learning Representations ({ICLR})},
  year      = {2025}
}

2024

[8]

Score-Optimal Diffusion Schedules

Christopher Williams, Andrew Campbell, Arnaud Doucet, Saifuddin Syed

Advances in Neural Information Processing Systems (NeurIPS), 2024

arXiv

@inproceedings{williams2024score,
  title     = {Score-Optimal Diffusion Schedules},
  author    = {Williams, Christopher and Campbell, Andrew and Doucet, Arnaud and Syed, Saifuddin},
  booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
  year      = {2024},
  note      = {arXiv:2412.07877}
}

[7]

autoMALA: Locally Adaptive Metropolis-Adjusted Langevin Algorithm

Miguel Biron-Lattes^*, Nikola Surjanovic^*, Saifuddin Syed, Trevor Campbell, Alexandre Bouchard-Côté

International Conference on Artificial Intelligence and Statistics (AISTATS), 2024

arXiv

@inproceedings{bironlattes2024automala,
  title     = {{autoMALA}: Locally adaptive {M}etropolis-adjusted {L}angevin algorithm},
  author    = {Biron-Lattes, Miguel and Surjanovic, Nikola and Syed, Saifuddin and Campbell, Trevor and Bouchard-C\^ot\'e, Alexandre},
  booktitle = {International Conference on Artificial Intelligence and Statistics ({AISTATS})},
  series    = {Proceedings of Machine Learning Research},
  volume    = {238},
  pages     = {4600--4608},
  year      = {2024},
  publisher = {PMLR},
  note      = {arXiv:2310.16782}
}

[6]

Uniform Ergodicity of Parallel Tempering with Efficient Local Exploration

Nikola Surjanovic, Saifuddin Syed, Alexandre Bouchard-Côté, Trevor Campbell

Under review · arXiv:2405.11384, 2024

arXiv

@misc{surjanovic2024uniform,
  title         = {Uniform Ergodicity of Parallel Tempering with Efficient Local Exploration},
  author        = {Surjanovic, Nikola and Syed, Saifuddin and Bouchard-C\^ot\'e, Alexandre and Campbell, Trevor},
  year          = {2024},
  eprint        = {2405.11384},
  archivePrefix = {arXiv},
  primaryClass  = {stat.CO}
}

2023

[5]

A Unified Framework for U-Net Design and Analysis

Fabian Falck^*, Christopher Williams^*, George Deligiannidis, Chris Holmes, Arnaud Doucet, Saifuddin Syed

Advances in Neural Information Processing Systems (NeurIPS), 2023

arXiv

@inproceedings{falck2023unified,
  title     = {A Unified Framework for {U}-{N}et Design and Analysis},
  author    = {Falck, Fabian and Williams, Christopher and Deligiannidis, George and Holmes, Chris and Doucet, Arnaud and Syed, Saifuddin},
  booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
  year      = {2023},
  note      = {arXiv:2305.19638}
}

[4]

Local Exchangeability

Trevor Campbell, Saifuddin Syed, Chiao-Yu Yang, Michael I. Jordan, Tamara Broderick

Bernoulli, 29(3):2084–2100, 2023

arXiv

@article{campbell2023local,
  title   = {Local Exchangeability},
  author  = {Campbell, Trevor and Syed, Saifuddin and Yang, Chiao-Yu and Jordan, Michael I. and Broderick, Tamara},
  journal = {Bernoulli},
  volume  = {29},
  number  = {3},
  pages   = {2084--2100},
  year    = {2023},
  note    = {arXiv:1906.09507}
}

2022

[3]

Parallel Tempering with a Variational Reference

Nikola Surjanovic, Saifuddin Syed, Trevor Campbell, Alexandre Bouchard-Côté

Advances in Neural Information Processing Systems (NeurIPS), 2022

arXiv

@inproceedings{surjanovic2022parallel,
  title     = {Parallel Tempering with a Variational Reference},
  author    = {Surjanovic, Nikola and Syed, Saifuddin and Campbell, Trevor and Bouchard-C\^ot\'e, Alexandre},
  booktitle = {Advances in Neural Information Processing Systems ({NeurIPS})},
  volume    = {35},
  pages     = {565--577},
  year      = {2022},
  note      = {arXiv:2206.00080}
}

[2]

Non-Reversible Parallel Tempering: A Scalable Highly Parallel MCMC Scheme

Saifuddin Syed, Alexandre Bouchard-Côté, George Deligiannidis, Arnaud Doucet

Journal of the Royal Statistical Society Series B, 84(2):321–350, 2022

arXiv DOI Code

@article{syed2022nonreversible,
  title   = {Non-Reversible Parallel Tempering: A Scalable Highly Parallel {MCMC} Scheme},
  author  = {Syed, Saifuddin and Bouchard-C\^ot\'e, Alexandre and Deligiannidis, George and Doucet, Arnaud},
  journal = {Journal of the Royal Statistical Society Series B: Statistical Methodology},
  volume  = {84},
  number  = {2},
  pages   = {321--350},
  year    = {2022},
  doi     = {10.1111/rssb.12464},
  note    = {arXiv:1905.02939}
}

2021

[1]

Parallel Tempering on Optimized Paths

Saifuddin Syed^*, Vittorio Romaniello^*, Trevor Campbell, Alexandre Bouchard-Côté

International Conference on Machine Learning (ICML), 2021

arXiv PMLR

@inproceedings{syed2021parallel,
  title     = {Parallel Tempering on Optimized Paths},
  author    = {Syed, Saifuddin and Romaniello, Vittorio and Campbell, Trevor and Bouchard-C\^ot\'e, Alexandre},
  booktitle = {Proceedings of the 38th International Conference on Machine Learning},
  series    = {Proceedings of Machine Learning Research},
  volume    = {139},
  pages     = {10033--10042},
  year      = {2021},
  publisher = {PMLR},
  note      = {arXiv:2102.07720}
}

^* equal contribution · ^† joint last author · All BibTeX · Full list on Google Scholar

Software

Pigeons.jl Julia

A package for sampling from intractable distributions, built on non-reversible parallel tempering. Pigeons runs unchanged from a single thread up to thousands of MPI-communicating machines, and is designed around parallelism invariance: the output for a given seed is identical no matter how many machines you run it on, which makes distributed randomised algorithms reproducible and testable.

Pigeons.jl

Research Group

Chris Williams

DPhil · Oxford · co-sup. A. Doucet, G. Deligiannidis

Peter Potaptchik

DPhil · Oxford · co-sup. G. Deligiannidis, Y. W. Teh

Leo Zhang

DPhil · Oxford · co-sup. R. Cornish, Y. W. Teh

Isaac Rankin

PhD · UBC · co-sup. C. Margossian

Teaching

STAT 547C Probability Theory