Jason Antic
Lead AI Engineer at Climavision
Greenfield, Massachusetts, United States
Summary
🤩
Rockstar🎓
Top SchoolJason Antic is a pragmatic lead AI engineer and founder with 16 years of experience turning research-grade deep learning into widely used products. He launched DeOldify in 2018, pioneering self-attention UNet architectures and practical GAN training techniques (NoGAN) that he co-developed with FastAI and showcased at Facebook F8, and his colorization/restoration models are deployed commercially on MyHeritage. Comfortable moving from research to production, he optimizes memory and performance in ML pipelines (notably improving diffusion notebooks for fp16 and efficiency) and builds lightweight, deployable models including CoreML for older iPhones. Jason pairs hands-on engineering—full-stack, test automation, and model productionization—with marketing and business management, having bootstrapped a two-person startup into a revenue-generating product. He thrives on creative problem solving with constrained resources and a disciplined approach to time and budget. Based in Greenfield, Massachusetts, he brings an uncommon mix of research depth and practical engineering that ships at scale.
16 years of coding experience
18 years of employment as a software developer
Bachelor of Science - BS Computer Science major minor in Mathematics, Bachelor of Science - BS Computer Science major minor in Mathematics at Penn State University
Stackoverflow
116reputation
53kreached
6answers
0questions
Github Skills (19)
pytorch10
python10
diffusion-models10
machine-learning10
stable-diffusion10
model-optimization10
cuda9
jupyter-notebook8
ai8
computer-vision7
syntax-highlighting6
database6
segment6
lua6
business-logic6
Programming languages (2)
Github contributions (5)
fastai/diffusion-nbs
Oct 2022 - Oct 2022
Getting started with diffusion
Role in this project:
ML Engineer
ML EngineerContributions:5 commits, 4 PRs, 10 comments in 8 days
Contributions summary:Jason primarily focused on optimizing memory efficiency within the diffusion model pipeline and modifying the example notebook. They removed unnecessary memory usage by replacing separate variable assignments and setting models to fp16. The user also updated example code to avoid safety filtering and fixed a typo to ensure the image generation loop worked correctly. Their contributions centered on improving the usability and performance of a diffusion model implementation.
deep-learningdiffusionfastai
jantic/KaggleCompetitions
Apr 2017 - Sep 2017
Contributions:45 commits, 40 pushes, 1 branch in 5 months
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.