Gabriel Rasskin
Software Engineer at Google
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolGabriel Rasskin is a software engineer based in San Francisco with seven years of experience building ML-driven systems and developer tools. Currently at Google working on open models, Keras, and Gemini-related projects, he has contributed production features to Keras—improving model compilation, learning-rate schedules, and adding several loss functions—and maintained its documentation and examples. His background spans applied research (NASA Goddard, IFIC) where he built fault-explanation time-series models and FPGA-quantized CNNs for real-time classification, showing strength at the intersection of ML research and engineering. Early work at Google included integrating fuzzing into TensorFlow and shipping a TF.js real-time depth network, demonstrating attention to robustness and web ML. He also has experience in mixed reality, conversational AI, and teaching, reflecting both product delivery and mentoring skills. Colleagues would note his blend of deep technical contribution to widely used open-source ML tooling and a knack for turning research prototypes into practical, production-ready systems.
7 years of coding experience
2 years of employment as a software developer
Bachelor of Science - BS, Computer Science, 3.55, Bachelor of Science - BS, Computer Science, 3.55 at Carnegie Mellon University
Github Skills (11)
neural-network10
keras10
machine-learning10
loss-functions10
deep-learning10
tensorflow10
python10
documentation10
data-science9
jax9
pytorch9
Programming languages (7)
Github contributions (5)
keras-team/keras-io
Feb 2023 - May 2024
Keras documentation, hosted live at keras.io
Role in this project:
Technical Writer
Technical WriterContributions:11 reviews, 14 PRs, 19 comments in 1 year 3 months
Contributions summary:Gabriel primarily focused on updating and improving the Keras documentation, as evidenced by the "Small spacing update" and "Knowledge distillation" commits. These commits involve modifications to existing documentation files within the `guides` and `examples` directories, suggesting a focus on clarity and completeness. Furthermore, the "Add movielens example" commit indicates the creation of new documentation and examples, extending the educational content related to Keras usage. The "Update timeseries/timeseries_classification_from_scratch example" and "Convert conv_lstm" commits further demonstrate the user's involvement in maintaining and converting existing documentation, contributing to the overall documentation quality.
deep-learninghostedmachine-learningtensorflowkeras
keras-team/keras
Oct 2022 - Dec 2022
Deep Learning for humans
Role in this project:ML Engineer
ML EngineerContributions:11 reviews, 4 commits, 14 PRs in 1 month
Contributions summary:Gabriel's contributions center on enhancing the Keras library, specifically focusing on model compilation, learning rate schedules, and loss functions. They added the ability to set the `jit_compile` property and implemented cosine decay with a warmup phase for learning rates. Additionally, the user introduced several new loss functions, including MeanSquaredLogarithmicError, MeanAbsolutePercentageError, Huber, LogCosh, and cosine similarity, along with associated testing.
pythondata-sciencedeep-learningneural-networksmachine-learning
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.