Gabriel Birnbaum
Lead Software Engineer at Can of Soup
San Francisco, California, United States
Summary
👤
Senior🎓
Top SchoolGabriel Birnbaum is a CEO, founder, and hands-on engineer with eight years of experience building AI-first products and scientific software from San Francisco. He co-founded Can of Soup and now leads Y/n, blending product design instincts (first product designer at Substack) with deep machine-learning engineering honed through leading Plasma.jl and significant contributions to NeuralPDE.jl and Zygote.jl. His open-source work focuses on physics-informed neural networks and automatic differentiation—fixing subtle tuple-dispatch issues in Zygote and implementing robust PINN solvers used by the SciML community. Gabriel pairs product-level impact (driving growth at Substack and press-covered consumer AI work) with rigorous numerical and symbolic engineering for PDEs and high-dimensional simulations. He studied computer science in Heidelberg and has participated in Y Combinator and Berkeley-affiliated programs, reflecting both startup grit and research collaboration. Notably, he combines storytelling-driven product visions with low-level numerical code, able to move between UX, model design, and compiler-like AD fixes.
8 years of coding experience
4 years of employment as a software developer
Grant awardee, Grant awardee at University of California, BerkeleyY CombinatorBachelor of Science, Computer Science, Bachelor of Science, Computer Science at Heidelberg UniversityFellow, Fellow at Steve Jobs Archive
Portuguese, Spanish, English, French, German, Hebrew, Swedish
Github Skills (10)
neural-network10
automatic-differentiation10
machine-learning10
differential-equations10
scim10
scientific-machine-learning10
julia10
gradient10
control-flow7
python4
Programming languages (8)
Github contributions (5)
SciML/NeuralPDE.jl
Jun 2021 - Jan 2022
Physics-Informed Neural Networks (PINN) Solvers of (Partial) Differential Equations for Scientific Machine Learning (SciML) accelerated simulation
Role in this project:
ML Engineer
ML EngineerContributions:21 reviews, 93 commits, 21 PRs in 6 months
Contributions summary:Gabriel primarily contributed to the development of physics-informed neural networks (PINNs) for solving partial differential equations. Their commits focused on implementing and refining core components for PINN solvers, including defining training strategies and adapting symbolic expressions. They addressed and resolved dimension mismatch errors, suggesting a focus on ensuring compatibility and robustness in the model's implementation and symbolic transformation of equations.
simulationscientific-machine-learningpartial-differential-equationsdifferentialode
FluxML/Zygote.jl
Sep 2021 - Oct 2021
21st century AD
Role in this project:Back-end Developer
Back-end DeveloperContributions:1 review, 7 commits, 3 PRs in 8 days
Contributions summary:Gabriel primarily focused on modifying and improving the handling of tuples within the Zygote.jl library, specifically related to automatic differentiation. Their contributions included adding and refining dispatch methods for tuples, fixing potential errors, and cleaning up the tuple handling logic. These changes aimed to ensure the correct and efficient processing of tuples within the automatic differentiation framework, impacting the library's ability to handle various data structures. Further contributions included merging branches and adding new tests related to map functions.
control-flowautomatic-differentiationmachine-learningjulia-compilerjulia
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.