Katy Crowson
Data And Solution Architect
Atlanta Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolKaty Crowson is a Data and Solution Architect with 13 years of experience driving AT&T’s network simplification and data transformation efforts, blending strategy, program leadership, and hands-on technical design. She specializes in designing scalable, secure Palantir-based data solutions and single-ontology integrations that turn disparate operational data into actionable insights and GenAI-enabled decisioning. Her background spans international consulting, large M&A integrations, and rebuilding business-IT relationships, backed by deep experience in project and organizational restructuring. An active open-source contributor, she has applied ML engineering skills to improve diffusion model training and evaluation in the notable k-diffusion PyTorch library and has strengthened numerical robustness in the colour-science project—an indication of both creative and precise technical instincts. Based in the Atlanta metro area, she pairs enterprise delivery discipline with a curiosity for generative AI and creative AI art.
12 years of coding experience
14 years of employment as a software developer
Vanderbilt UniversityLamar University
Stackoverflow
1reputation
0reached
0answers
0questions
Github Skills (9)
diffusion-models10
pytorch10
machine-learning10
colors10
evaluation10
python10
metric10
numpy10
documentation7
Programming languages (11)
Github contributions (5)
crowsonkb/k-diffusion
Jun 2022 - Jan 2023
Karras et al. (2022) diffusion models for PyTorch
Role in this project:
ML Engineer
ML EngineerContributions:1 review, 149 commits, 10 PRs in 6 months
Contributions summary:Katy primarily focused on improving the training and evaluation of diffusion models within the `k-diffusion` library. Their contributions included adding automatic FID/KID evaluation metrics, logging these metrics, and integrating a CLIP feature extractor for better evaluation. They also made changes to training scripts, including adjusting the learning rate and incorporating a stochastic sampling method, and the addition of code related to the exponential noise schedule, all of which improved model training and evaluation.
pytorchdeep-learningdiffusiondiffusion-modelspytorch-lightning
colour-science/colour
May 2017 - May 2017
Colour Science for Python
Role in this project:Back-end Developer
Back-end DeveloperContributions:5 commits, 7 PRs, 6 comments in 5 days
Contributions summary:Katy's contributions primarily focus on improving the robustness and accuracy of the `colour-science/colour` library. They addressed potential numerical issues by handling negative values in the `post_adaptation_non_linear_response_compression_forward()` function and preventing divide-by-zero warnings in `CIECAM02` calculations. Additionally, the user corrected documentation, specifically fixing the publication year for a referenced paper.
colourspacesdatapythoncolour-spacescolor-space
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.