Keon Kim
Co-Founder at Om Labs
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolKeon Kim is a San Francisco–based co-founder and software engineer with 11 years of experience building backend systems and production ML tooling. He previously led backend work at Uber on driver pricing and incentive products, and now builds AI agents at Om Labs after selling wallet infrastructure IP to Silence Labs. Keon combines practical ML expertise—demonstrated by open-source implementations of DQN/DDQN, seq2seq with attention, and reinforcement-learning examples—with robust backend skills in C++ and Python. He’s an active contributor to mlpack and clean educational repos that reflect a focus on reproducible, minimal implementations for learning and production. An angel investor in ambitious deep-tech and space startups, he brings a founder’s perspective to product and engineering trade-offs. Collectedly, his profile blends hands‑on ML engineering, systems design, and early-stage investing insight.
11 years of coding experience
2 years of employment as a software developer
Bachelor of Arts (BA) Computer Science, Bachelor of Arts (BA) Computer Science at New York University
Stackoverflow
750reputation
115kreached
7answers
36questions
Github Skills (51)
algorithms10
pytorch10
machine-translation10
c-language10
python10
encoder10
attention-mechanism10
command-line-interface10
machine-learning10
decode10
seq2seq10
reinforcement-learning10
data-structure10
keras10
dqn10
Programming languages (12)
Github contributions (5)
rlcode/reinforcement-learning
Jan 2017 - Jul 2017
Minimal and Clean Reinforcement Learning Examples
Role in this project:
ML Engineer
ML EngineerContributions:59 commits, 39 PRs, 57 pushes in 5 months
Contributions summary:Keon primarily contributed to reinforcement learning examples, focusing on implementing and updating deep Q-network (DQN) algorithms for the CartPole environment. They modified existing code related to DQN, Keras, and the CartPole environment, and also created/updated files for the Keras CartPole DQN, REINFORCE, and other related files. The contributions demonstrate the user's focus on practical applications of reinforcement learning techniques.
a3cactor-criticdeep-learningreinforcement-learningdeep-reinforcement-learning
keon/seq2seq
Dec 2017 - Jul 2020
Minimal Seq2Seq model with Attention for Neural Machine Translation in PyTorch
Role in this project:ML Engineer
ML EngineerContributions:21 commits, 9 PRs, 30 pushes in 2 years 8 months
Contributions summary:Keon primarily contributed to the development of a Seq2Seq model for neural machine translation using PyTorch. Their commits involve implementing core model components such as the Encoder, Decoder, and Attention mechanisms. The user also worked on training the model, including defining the training loop, loss calculation, and optimization steps. They also optimized memory usage and refactored code.
pytorchseq2seqmachine-translationseq2seq-modeltranslation
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.