Gerardo Duran-martin
Postdoctoral Researcher at University of Oxford
United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolGerardo Duran-martin is a quantitative researcher with 11 years of experience blending statistical machine learning and state-space modeling, currently at Tower Research Capital after a postdoc at the University of Oxford. He holds a PhD in Mathematical Sciences and has a track record of impactful internships at Man AHL and contributions to high-profile open-source projects like Kevin Murphy’s probml and Dynamax, where he implemented and refactored EKF/UKF and probabilistic demos. Skilled in JAX and probabilistic methods, he has translated classical algorithms into modern ML tooling through GSoC work for TensorFlow/JAX. Colleagues value his ability to move between rigorous theory—hierarchical Bayesian models, Gibbs sampling, Kalman filters—and production-ready code that surfaces subtle model behaviors such as GMM singularities.
11 years of coding experience
2 years of employment as a software developer
Bachelor’s Degree, Actuarial Science, Bachelor’s Degree, Actuarial Science at Universidad MaristaDoctor of Philosophy - PhD, Mathematical Sciences, Doctor of Philosophy - PhD, Mathematical Sciences at Queen Mary University of London
English, Spanish
Stackoverflow
101reputation
67reached
0answers
1question
Github Skills (12)
data-visualizations10
bayesian-statistics10
machine-learning10
data-visualization10
data-visualisation10
bayesian10
jax10
python10
kalman-filter10
scipy8
probabilistic-programming7
oop6
Programming languages (12)
Github contributions (5)
probml/dynamax
Nov 2022 - Jan 2023
State Space Models library in JAX
Role in this project:
ML Engineer
ML EngineerContributions:2 reviews, 71 commits, 3 PRs in 2 months
Contributions summary:Gerardo refactored and extended the Extended Kalman Filter (EKF) implementation within the `dynamax` library. They added functionality to optionally return historical filtered means, covariances, and marginal log-likelihoods, contributing to the development of state space models. Additionally, the user made changes to the Unscented Kalman Filter (UKF) to return historical filtered means and covariances and updated notebooks related to EKF and UKF for spiral data.
pythonstate-space-modelsdynamical-systemskalman-filterstate-space
probml/pyprobml
Jan 2021 - Jul 2022
Python code for "Probabilistic Machine learning" book by Kevin Murphy
Role in this project:Data Scientist
Data ScientistContributions:18 reviews, 61 commits, 59 PRs in 1 year 6 months
Contributions summary:Gerardo contributed to the project by implementing and visualizing Gaussian Mixture Models (GMMs) for the Old Faithful dataset, as well as illustrating singularities in GMMs. They also developed examples of Parzen window density estimation and created a demo visualizing the geometry of ridge regression. Further contributions include examples of Gibbs sampling, Kalman filtering, and a hierarchical Bayesian model, demonstrating proficiency in machine learning and statistical modeling.
pythonpmlflaxcolabpymc3
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.