Fariz Rahman
Co-Founder at Stealth Startup
Kochi, Kerala, India
Summary
🤩
Rockstar🎓
Top SchoolFariz Rahman is a deep learning engineer and co-founder with 10 years of experience building ML systems and developer tools from Kochi, India. He has a strong open-source footprint—contributing to Keras and related projects and authoring libraries like recurrentshop and seq2seq that simplify RNN construction and training for practitioners. His industry experience spans startups and platforms (Weights & Biases, Activeloop, Deeplearning4j) where he worked on model import, backend integrations, and tooling to make complex models production-ready. Fariz focuses on democratizing deep learning by designing simple, usable APIs over advanced techniques so small teams and amateur developers can adopt them. He combines research-grade RNN expertise (teacher forcing, custom RNN cells) with practical backend and deployment work, from importers to data storage. Known for polishing code quality and interoperability, he bridges research, engineering, and product to accelerate ML adoption.
10 years of coding experience
8 years of employment as a software developer
+2 Computer Science, +2 Computer Science at MES International SchoolBachelor of Technology (B.Tech.) Computer Science, Bachelor of Technology (B.Tech.) Computer Science at Cochin University of Science and Technology
Malayalam, Hindi, Arabic, English
Github Skills (32)
back-end-development10
python10
data-science10
machine-learning10
recurrent-neural-networks10
java10
javas10
keras10
deeplearning-ai10
lstm10
deep-learning10
api10
tensorflow10
natural-language-processing10
neural-network10
Programming languages (8)
Github contributions (5)
farizrahman4u/recurrentshop
Sep 2016 - Jul 2018
Framework for building complex recurrent neural networks with Keras
Role in this project:
ML Engineer
ML EngineerContributions:155 commits, 35 PRs, 138 pushes in 1 year 9 months
Contributions summary:Fariz primarily contributed to the development of a framework for building recurrent neural networks, recurrentshop. Their work focused on defining the core structure of the framework, including RNNCell and RecurrentContainer classes that allow users to build complex recurrent neural networks with Keras. They implemented various RNN cell types such as SimpleRNNCell, GRUCell, and LSTMCell, and added features like teacher forcing and readout functionality, demonstrating their expertise in recurrent neural networks and deep learning model development. The user also added example code and made improvements to the internal mechanisms to ensure proper functionality.
deep-learningrecurrent-neural-networksneural-networksneural-networktensorflow
farizrahman4u/seq2seq
Nov 2015 - Aug 2022
Sequence to Sequence Learning with Keras
Role in this project:ML Engineer
ML EngineerContributions:241 commits, 33 PRs, 229 pushes in 6 years 10 months
Contributions summary:Fariz primarily worked on implementing and refining a sequence-to-sequence model for natural language processing tasks, as evidenced by the creation of LSTM-based encoder and decoder layers. They refactored the code, renamed files, and integrated the new components to build a complete model. Additionally, the user added features like teacher forcing to improve model training.
sequencedeep-learningtensorflowsequence-learningkeras
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.