Daniel Jiang
Research Scientist at Google DeepMind
New York, New York, United States
Summary
🤩
Rockstar🎓
Top SchoolDaniel Jiang is a Research Scientist at Meta with a decade of experience bridging rigorous academic research and production-grade machine learning systems. He holds a Ph.D. in Operations Research and Financial Engineering from Princeton and dual highest-distinction B.S. degrees in Computer Engineering and Mathematics from Purdue, reflecting deep quantitative foundations. At Meta he applies Bayesian optimization and probabilistic modeling expertise—evidenced by substantive contributions to prominent open-source projects like BoTorch and Ax, including implementing a MultivariateNormal quasi-Monte Carlo sampler and enhancements to learned cost modeling. Comfortable in both research and backend engineering roles, he pairs careful unit-tested implementations with applied experimentation thinking. Based in New York, he maintains a public research presence and often works at the intersection of statistical methods and scalable ML tooling.
10 years of coding experience
10 years of employment as a software developer
B.S. Electrical and Computer Engineering B.S. Mathematics, B.S. Electrical and Computer Engineering B.S. Mathematics at Purdue UniversityWest Lafayette Jr/Sr High SchoolDoctor of Philosophy (Ph.D.) Operations Research and Financial Engineering, Doctor of Philosophy (Ph.D.) Operations Research and Financial Engineering at Princeton University
Github Skills (11)
unit-testing10
pytorch10
machine-learning10
bayesian10
optimisation10
python10
optimization10
data-science9
linear-algebra9
test-driven-design8
gpytorch8
Programming languages (3)
Github contributions (5)
facebook/Ax
Dec 2019 - Sep 2022
Adaptive Experimentation Platform
Role in this project:
Back-end Developer & Data Scientist
Back-end Developer & Data ScientistContributions:62 commits, 34 PRs, 4 comments in 2 years 9 months
Contributions summary:Daniel contributed to the development of the `ax` library, focusing on enhancing its capabilities for cost modeling and experimentation. They implemented the passing of `metric_names` to facilitate the use of learned cost models that require log transforms and other special modeling techniques. Their work included modifications to test files to incorporate the new functionality, indicating a focus on ensuring the robustness of their changes. The user's contributions involved code changes related to BoTorch models, demonstrating skills relevant to machine learning and Bayesian optimization.
experimentationadaptivesimulationadaptive-experimentation-platformexperimentation-platform
meta-pytorch/botorch
Jan 2019 - May 2022
Bayesian optimization in PyTorch
Role in this project:ML Engineer
ML EngineerContributions:1 release, 67 commits, 20 PRs in 3 years 4 months
Contributions summary:Daniel's primary contributions involve implementing and testing a new Multivariate Normal qMC sampling capability within the BoTorch framework. This included adding a `MultivariateNormalQMCEngine` class for drawing samples from multivariate normal distributions using quasi-Monte Carlo methods. The user also incorporated unit tests to validate the new functionality, including tests for various scenarios such as non-positive definite covariance matrices, different dimensions, inverse transforms, and seeded random number generators. Furthermore, the user integrated the ability to sample from degenerate MVNs using eigendecomposition.
bayesianpytorch
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.