Si-Qi LIU
Staff Research Engineer at DeepMind
London, England, United Kingdom
Summary
🤩
Rockstar🎓
Top SchoolSi-Qi LIU is a Staff Research Engineer at DeepMind with 13 years of experience building research-grade systems for reinforcement learning and physics-based simulation. He progressed through multiple senior research engineering roles at DeepMind after internships at Google and Google Research, combining strong software engineering with ML research instincts. His open-source contributions include substantive backend work on DeepMind/OpenSpiel—adding egocentric observations and a max-entropy best-response—and enhancements to dm_control’s soccer environment such as reward shaping, tracking cameras, and multi-turn tasks. Trained across top institutions (UCL PhD in AI & Neuroscience, MSc Oxford, MEng CentraleSupélec), he bridges rigorous academic methods with production-quality C++/Python implementations. Colleagues rely on him for turning game-theoretic and simulation ideas into robust, extensible code rather than toy prototypes. He brings a rare mix of deep theoretical grounding and pragmatic engineering that accelerates research-to-product workflows.
13 years of coding experience
6 years of employment as a software developer
Master of Engineering (MEng) Diplôme d'ingénieur Computer and Information Sciences General, Master of Engineering (MEng) Diplôme d'ingénieur Computer and Information Sciences General at CentraleSupélecUniversity College LondonMaster of Science (MSc) Computer Science, Master of Science (MSc) Computer Science at University of OxfordChinese High School Degree science track, Chinese High School Degree science track at The High School Affiliated to Renmin University of ChinaMaster’s Degree in Engineering/Diplôme d'ingénieur Computer Science, Master’s Degree in Engineering/Diplôme d'ingénieur Computer Science at Université de Technologie de Compiègne (UTC)
Chinese, English, French, Japanese, German
Stackoverflow
165reputation
132kreached
3answers
4questions
Github Skills (28)
artificial-intelligence10
c-language10
python10
coregames10
machine-learning10
reinforcement-learning10
videogames10
deeplearning-ai10
deep-learning10
cprogramming-language10
simulation9
algorithms9
multi-agent9
simulations9
physics9
Programming languages (5)
Github contributions (5)
google-deepmind/dm_control
Feb 2019 - Jun 2021
Google DeepMind's software stack for physics-based simulation and Reinforcement Learning environments, using MuJoCo.
Role in this project:
ML Engineer
ML EngineerContributions:25 commits, 1 push, 3 comments in 2 years 3 months
Contributions summary:Si-Qi primarily contributed to the development of the soccer environment within the dm_control library, focusing on enhancing its functionality and realism for reinforcement learning research. They implemented environment rewards based on goals scored and conceded, improving the environment's usability for training agents. Furthermore, the user added features, such as tracking cameras and forward velocity observation for players, which provide more information for training and evaluation. They also introduced a multi-turn variant of the task, facilitating more complex training scenarios.
simulationreinforcementsoftware-stackphysics-based-simulationsimulation-based
google-deepmind/open_spiel
Nov 2021 - Sep 2022
OpenSpiel is a collection of environments and algorithms for research in general reinforcement learning and search/planning in games.
Role in this project:Back-end Developer
Back-end DeveloperContributions:4 commits in 9 months
Contributions summary:Si-Qi primarily contributed to the `open_spiel` repository by enhancing the Goofspiel game implementation. They introduced an option for egocentric observations, modifying the game's observer to include this feature. Furthermore, the user implemented improvements to the best response algorithm, providing a max-entropy best response option. In addition, they also added new variants and fixed typo errors in the Python and C++ code.
cppmultiagentgamespythondatamining
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.