Danijar Hafner
San Francisco, California, United States
Summary
🤩
Rockstar🎓
Top SchoolDanijar Hafner is a research scientist and machine learning engineer with 12 years of experience building world-model-based AI that makes autonomous decisions, currently located in San Francisco. He has a deep research and engineering track record at Google and DeepMind, progressing from multiple student and intern roles to senior and staff research scientist positions. His open-source work includes influential contributions to model-based RL projects such as DreamerV2 and Planet, improving benchmarks, visualization, and environment support. He blends strong software engineering (full-stack tooling like the handout project) with rigorous PhD-level research from the University of Toronto and visiting work at UC Berkeley. Known for practical experimentation—adding GIF summaries to TensorBoard and integrating Dopamine baselines—he focuses on temporal abstraction, curiosity-driven exploration, and latent dynamics. Quiet about LinkedIn, he prefers email contact and lets his code and papers speak for his work.
12 years of coding experience
10 years of employment as a software developer
Visiting PhD Student, Visiting PhD Student at University of California, BerkeleyBSc, IT Systems Engineering, BSc, IT Systems Engineering at Hasso Plattner InstitutePhD, Computer Science, PhD, Computer Science at University of TorontoUniversity College London
Stackoverflow
33,334reputation
5.1mreached
108answers
369questions
css
top-5%
machine-learning
top-5%
tensorflow
top-1%
jquery
top-5%
keras
top-5%
printf
top-5%
Github Skills (44)
algorithm10
markdown10
algorithms10
python10
model-driven10
machine-learning10
document-generation10
reinforcement-learning10
implement10
model-building10
deep-learning10
markdown-it10
tensorflow10
modeling10
model-driven-development10
Programming languages (25)
Github contributions (5)
danijar/handout
Aug 2018 - Apr 2020
Turn Python scripts into handouts with Markdown and figures
Role in this project:
Full-stack Developer
Full-stack DeveloperContributions:45 commits, 8 PRs, 52 pushes in 1 year 7 months
Contributions summary:Danijar primarily developed the `handout` project, a tool to turn Python scripts into handouts. Their work involved implementing a Python-based framework, including features for incorporating Markdown text, images, and videos. They refactored the code into separate files, added code highlighting using JavaScript, and improved the user experience with print styles and other enhancements.
markdownpython-scriptspythondata-sciencenotebook
google-research/batch-ppo
Aug 2017 - Jan 2019
Efficient Batched Reinforcement Learning in TensorFlow
Role in this project:ML Engineer
ML EngineerContributions:5 releases, 51 commits, 46 PRs in 1 year 4 months
Contributions summary:Danijar primarily contributed to the implementation and refinement of the PPO (Proximal Policy Optimization) reinforcement learning algorithm within a TensorFlow environment. Their work includes updating the setup file, refactoring algorithm functions, and modifying network interfaces. Furthermore, they focused on improving the training and evaluation scoring mechanisms, enhancing the overall performance and reporting capabilities of the PPO implementation. The user also addressed issues related to state management and memory operations.
reinforcement-learningtensorflowmulti-processingartificial-intelligencepython
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.