Erik Welch
OSS Developer at NVIDIA
Austin, Texas, United States
Summary
🤩
Rockstar🎓
Top SchoolErik Welch is an OSS-focused back-end developer with 12 years of experience building and hardening high-performance Python libraries for graph analytics and parallel computing. Currently at NVIDIA after eight years at Anaconda, he contributes to prominent projects like Dask, NetworkX, cuGraph, and Toolz, improving task scheduling, graph algorithms, memory usage, and Python compatibility. His work balances new feature development (e.g., apply/compose_left in Toolz) with pragmatic refactors such as versioneer adoption, non-recursive graph traversal, and reducing redundant data storage for better performance. He pairs strong testing and QA instincts—relaxing numerical comparisons and adding algorithm tests—with hands-on optimization of import times and API ergonomics. Based in Austin, he brings an engineering-physics background to large-scale open-source engineering, often surfacing subtle correctness and maintainability wins that pay off in production.
13 years of coding experience
7 years of employment as a software developer
Engineering Physics, Engineering Physics at The Ohio State University
Github Skills (36)
scheduled-tasks10
graph-algorithms10
lib10
pytest10
python10
testing10
dataframes10
graph10
pandas10
graph-analysis10
dataframe10
numpy10
scheduler10
scheduling10
sched10
Programming languages (13)
Github contributions (5)
pytoolz/toolz
Oct 2019 - Nov 2022
A functional standard library for Python.
Role in this project:
Back-end Developer
Back-end DeveloperContributions:6 releases, 3 reviews, 57 commits in 3 years 1 month
Contributions summary:Erik primarily contributed to the maintenance and improvement of the `toolz` library. They added new functionalities and updated existing ones, specifically introducing `apply`, `compose_left` and refactoring code for better organization. The user also refactored the project to use `versioneer` for version management, updated supported Python versions, and addressed a Python 3.11 compatibility issue. Finally, they replaced deprecated terms and optimized import times for faster operation.
python
dask/dask
Jan 2015 - Feb 2023
Parallel computing with task scheduling
Role in this project:Back-end Developer
Back-end DeveloperContributions:17 reviews, 30 PRs, 2 pushes in 8 years 2 months
Contributions summary:Erik primarily contributed to the core functionality of the Dask library, enhancing its parallel computing capabilities. Their work involved implementing the `cull` function to remove unnecessary tasks and improving graph traversal with the addition of `isdag` and a non-recursive `get` function to handle large graphs. The commits demonstrate a focus on optimizing the performance and reliability of Dask's task scheduling and graph execution. The user also improved the `fuse` functionality.
parallel-computingdaskpythonpydatanumpy
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.