Mihail Dumitrescu
Senior Software Engineer at Google
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolMihail Dumitrescu is a Senior Software Engineer based in London with 11 years of professional experience building high-performance systems at Google after early internships at Palantir and Google. He holds a B.S. in Computer Science from Caltech and combines strong foundations in algorithms with practical expertise in optimizing complex ML workloads. As an ML engineer contributor to the popular InvokeAI project, he focused on refactoring attention mechanisms, reducing VRAM footprint and improving CPU/precision performance to make Stable Diffusion workflows more efficient. Known for pairing deep technical rigor with pragmatic performance engineering, he thrives on squeezing latency and memory gains out of real-world models and production services.
11 years of coding experience
1 year of employment as a software developer
Bachelor of Science (B.S.), Computer Science, Bachelor of Science (B.S.), Computer Science at California Institute of Technology
Romanian, Spanish
Github Skills (8)
pytorch10
artificial-intelligence10
stable-diffusion10
image-generation10
python10
cuda9
computer-vision8
mlops7
Programming languages (7)
Github contributions (5)
invoke-ai/InvokeAI
Sep 2022 - Sep 2022
Invoke is a leading creative engine for Stable Diffusion models, empowering professionals, artists, and enthusiasts to generate and create visual media using the latest AI-driven technologies. The solution offers an industry leading WebUI, and serves as the foundation for multiple commercial products.
Role in this project:
ML Engineer
ML EngineerContributions:12 reviews, 11 commits, 15 PRs in 13 days
Contributions summary:Mihail primarily focused on optimizing and refactoring code related to the Stable Diffusion model, including speed improvements and VRAM management. They addressed issues related to CPU utilization and precision settings, making the application more efficient. The contributions also involved refactoring and optimization of attention mechanisms within the model architecture, along with fixing potential memory-related issues.
serveswebuioutpaintingdiffusionlatent-diffusion
mh-dm/Marlin
Jun 2021 - Dec 2024
Fork of Marlin to submit PRs
Contributions:52 pushes, 46 branches in 3 years 5 months
vendorfirmwarereprapavratmel
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.