My name is Steve Bronder and I’m a Developer specializing in Computational Statistics

My C++ compiles pretty often which is cool. I enjoy working on Open Source projects

Know more Let’s Have Coffee

Steve Bronder

Stan Developer,

Columbia University

Steve Bronder is a developer for the Stan programming language. His research interest include high performance computing, Bayesian statistics, and open source development. After finishing his masters in Quantitative Methods In the Social Sciences at Columbia University, he began working as a Data Scientist at Capital One. There he focused on Risk Modeling for collateralized debt obligations and teaching business colleagues how to automate the boring parts of their jobs.

Steve Bronder is available for Consulting services through Varkara LLC, please send an email to varkara@stevebronder.com for rates.

Steve is originally from Pittsburgh, Pennsylvania and now lives in New York City. He has a cat named Hopscotch and enjoys cooking, playing Dungeons & Dragons / Magic: The Gathing, and working on other open source projects.

Download Resume

Skills

I AM GOOD AT THESE AND OTHER THINGS

C++

R

Python

Forecasting

Experience

Some Cool and Fun Things I’ve done in the past

Stan

Developer

Feb 2020 – Present New York City

Building Language Features For:

Sparse Matrices
Automatic Differentiation
GPU Computing
Parallelization

Capital One

Data Scientist

May 2018 – Feb 2020 New York City

Developed risk models for CDO style contracts
Built Machine Learning Models to Impute Moody’s Risk Ratings
Automated Manual Reporting

Zurich Insurance

Predictive Analyst Intern

May 2016 – Aug 2016 New York City

Wrote Predictive Models for Losses on Corporate Auto Contracts
Constructed new validation metrics for assesing loss model performance
Created Web Apps for managers to track timelines for analytics projects

Publications

Stuff that I liked so much I wrote a whole lot about them

GPU Support for The Stan Math Library

Steve Bronder, Rok Češnovar, Davor Sluga, Jure Demšar, Tadej Ciglarič, Sean Talts, Erik Štrumbelj

July 2019 Arvix

Building A GPU backend for the Stan Math Library.

PDF

Approximations using GPUs for Large Traveling Salesman Problems

Steve Bronder, Haoyan Min

April 2019 Columbia University Electrical Engineering Student Symposium

You can do a ton of Monte Carlo samples at the same time to get good traveling salesman soutions

PDF

Tuning machine learning and time series models using ML methodologies

Steve Bronder

September 2016 R In Finance 2017

Tuning machine learning models for forecasting projects

PDF