Azamat Berdyshev
Founding Machine Learning Engineer at Godela (YC X25)
San Francisco, California, United States
Summary
👤
Senior🎓
Top SchoolAzamat Berdyshev is a Founding Machine Learning Engineer with a decade of experience building and shipping core ML models and infrastructure across startups and research settings. He has led end-to-end model delivery as Principal ML Engineer at Symbolic Mind and now shapes product and architecture as employee #1 at Godela (YC X25) from San Francisco. His background blends rigorous applied math and ML (MASc, U of T) with practical systems work—from quant derivatives tooling at RBC to production ML at RelationalAI and Sophi.io. An active open-source contributor, he improved automatic differentiation in the widely used FluxML/Zygote.jl library by refining adjoints and test coverage, showing deep understanding of gradient math and correctness. Colleagues describe him as someone who prefers complex problems and autonomy, pairing sharp technical judgment with a team-first, impact-driven mindset. He’s motivated not just by elegant solutions but by making a measurable difference and looking out for others along the way.
10 years of coding experience
8 years of employment as a software developer
Master of Applied Science (MASc) Machine Learning, Master of Applied Science (MASc) Machine Learning at University of TorontoHigh School Diploma, High School Diploma at National School of Physics & Math (FIZMAT)
English, qazaq, Russian
Github Skills (5)
automatic-differentiation10
machine-learning10
julia10
gradient10
testing8
Programming languages (6)
Github contributions (5)
FluxML/Zygote.jl
Jan 2020 - Feb 2020
21st century AD
Role in this project:
ML Engineer
ML EngineerContributions:7 commits, 8 PRs, 53 comments in 13 days
Contributions summary:Azamat focused on enhancing the automatic differentiation capabilities of the `zygote.jl` library, a core component of the repository. Their contributions primarily involved adding and refactoring adjoints for functions like `repeat`, and applying `@nograd` to control certain functions, directly impacting gradient calculations within the Julia-based machine learning framework. These changes demonstrate an understanding of the mathematical operations at the heart of automatic differentiation. Further contributions included adding tests to ensure the correctness of gradient calculations.
control-flowautomatic-differentiationmachine-learningjulia-compilerjulia
AzamatB/WaveNet.jl
Mar 2020 - Feb 2022
Contributions:9 commits, 2 pushes in 1 year 11 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.