Parsiad Azimzadeh
Vice President at PDT Partners
United States
Summary
👤
Senior🎓
Top SchoolParsiad Azimzadeh is a Vice President and machine learning engineer with 12 years of experience applying mathematical rigor to production-grade probabilistic and optimization systems. Currently at PDT Partners, he brings a research-rooted perspective from roles at Google and the University of Michigan and holds advanced degrees in mathematics from the University of Waterloo. His open-source contributions include improving ODE solvers and adjoint gradients in TensorFlow Probability and refining Pareto-front algorithms in Optuna—work that strengthens gradient-based training and multi-objective optimization in widely used ML toolkits. Known for blending deep theoretical insight with pragmatic engineering, he often focuses on edge cases and numerical robustness that quietly make models more reliable in practice.
12 years of coding experience
1 year of employment as a software developer
Master of Mathematics, Master of Mathematics at University of WaterlooBachelor's degree, Bachelor's degree at Simon Fraser University
English, farsi (persian)
Github Skills (21)
algorithm10
algorithms10
probabilistic-programming10
ode10
python10
solver10
machine-learning10
data-structure10
hyperparameter-optimization10
tensorflow10
ode-solver10
data-structures10
odeint10
testing9
statistics9
Programming languages (8)
Github contributions (5)
tensorflow/probability
May 2019 - Aug 2019
Probabilistic reasoning and statistical analysis in TensorFlow
Role in this project:
ML Engineer
ML EngineerContributions:5 commits in 3 months
Contributions summary:Parsiad focused on enhancing the TensorFlow Probability library's ODE solvers. Their contributions included fixing validation steps in the BDF solver, refactoring ODE solvers, and implementing the Adjoint Method for gradients with respect to the initial state. These changes improved the robustness and functionality of the solvers, particularly for gradient calculations, which is critical for training and analyzing probabilistic models.
statisticspythonprobabilistic-reasoningdata-sciencedeep-learning
optuna/optuna
Mar 2021 - Mar 2021
A hyperparameter optimization framework
Role in this project:ML Engineer
ML EngineerContributions:3 reviews, 5 commits, 1 PR in 5 days
Contributions summary:Parsiad contributed to the optimization of the multi-objective Pareto front algorithm within the Optuna framework. Their work included implementing a log-linear algorithm for 2D Pareto fronts, addressing edge cases, and integrating the use of the `_dominates` function. Further improvements involved adding a 3D Pareto front test case and refactoring the 2D Pareto front algorithm.
pythonoptimization-frameworkparallelhyperparameteroptimization
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.