Bryan Lyon
CTO at faceswap.dev
Los Angeles Metropolitan Area United States
Summary
🤩
Rockstar🎓
Top SchoolBryan Lyon is a CTO and seasoned AI/ML leader with 11 years of experience building production-grade deep learning systems and leading engineering teams from Los Angeles. Trained originally as an ESL teacher with an MA in TESOL and a BA in Linguistics, he combines world-class expertise in grammar, language acquisition, and NLP with hands-on machine learning engineering. He founded and continues to drive faceswap.dev and contributes to the prominent deepfakes/faceswap open-source project, where he implemented multi-GPU support to optimize GAN and related models for scalable training. As CTO of DeepMake and former CTO of Doppel AI, he has a track record of translating research-grade models into deployable products. His background as a US Army cryptologic linguist gives him rare operational experience applying language analysis under real-world constraints. Bryan bridges linguistics, ML research, and engineering leadership to ship impactful, resource-efficient AI systems.
11 years of coding experience
7 years of employment as a software developer
Master of Arts (M.A.) Teaching English to Speakers of Other Languages (TESOL), Master of Arts (M.A.) Teaching English to Speakers of Other Languages (TESOL) at Minnesota State University, MankatoThe University of UtahAssociate of Arts (A.A.) English, Associate of Arts (A.A.) English at Salt Lake Community College
English, Japanese, arabic (modern standard)
Github Skills (11)
neural-network10
keras10
computer-vision10
machine-learning10
deep-learning10
cgan10
tensorflow10
python10
dcgan10
model-optimization10
autoencoder9
Programming languages (7)
Github contributions (5)
deepfakes/faceswap
Mar 2018 - May 2020
Deepfakes Software For All
Role in this project:
ML Engineer
ML EngineerContributions:21 commits, 38 PRs, 18 pushes in 2 years 2 months
Contributions summary:Bryan primarily focused on adding multi-GPU support to the deepfakes software, specifically within the GAN, Original, IAE, and LowMem models. This involved modifying existing code, refactoring model structures, and integrating multi-GPU functionality using `keras.utils.multi_gpu_model`. The commits demonstrate expertise in adapting deep learning models for improved performance and resource utilization. The changes include updates to model loading and saving, reflecting a focus on optimizing the existing codebase for efficient training on multiple GPUs.
deepfakesdeep-learningneural-netsmachine-learningface-swap
DeepMakeStudio/DeepMake
Nov 2023 - Sep 2024
DeepMake
Contributions:5 releases, 6 PRs, 139 pushes in 9 months
after-effectsgenerative-aispecial-effectsvfxvisual-effects
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.