Email: leonid.shamis@gmail.com
Phone: (916) 402-0199
GitHub: lshamis alephzero
Staff Software Engineer with expertise in machine learning, robotics, NLP, and distributed systems, with a proven track record of developing and deploying advanced AI systems.
Programming Languages: C/C++, Python, JavaScript, Go, Rust, Java, C#
Machine Learning & AI: PyTorch, TensorFlow, SIMD, NLP, Robotics
Systems & Infrastructure: Linux, Containers (Docker/Apptainer/Enroot), SLURM, Ray, Distributed Systems
Development Tools: Git, Bash, Cross-Language Integration (JNI, PyBind, CGo), Performance Optimization
Led Chameleon (multi-modal foundation model) inference optimization, demo infrastructure, and open-source release with HuggingFace.
Developed ALMA, a low-annotation LLM alignment method enhancing controllability.
Primary maintainer of FAIR's new LLM training codebase to support future model development. Hopefully to be open sourced soon.
Managed research teams for mobility, manipulation, and home robotics.
Led development of the Meta Robotics Platform (MRP) for scalable robotics research.
Led teams developing autonomous vehicle motion planning, SLAM, and orchestration.
Designed scalable motion planning and perception systems for autonomous forklifts.
Founding engineer. Built unified robotic data transfer and orchestration system.
Led motion planning team, designing simulation tools and semantic map representations. Developed algorithms for route planning and vehicle dynamics.
Researched reinforcement learning and deep learning methods for large-scale machine learning applications.
Developed control software for multi-agent robotic systems and designed safety mechanisms for experimental hardware.
Built NLP models for sentiment analysis, QA, and personalized search. Designed systems to extract testimonials and opinions from Play Store comments.
Built large-scale data analysis pipelines and A/B testing infrastructure for Google's search ranking. Developed automated deployment, monitoring, and traffic prediction systems.
Constraint Based Framework for Optimal k-Clustering.
Master's Thesis. University of California, Davis, 2010.
A SAT-based Framework for Efficient Constrained Clustering.
Davidson, Ian, Ravi, S. S. and Shamis, Leonid. SDM, 2010.
Chameleon: Mixed-Modal Early-Fusion Foundation Models.
Chameleon Team. arXiv:2405.09818, May 2024.
Transfusion: Predict the Next Token and Diffuse Images with One Multi-Modal Model.
Chunting Zhou, Lili Yu, Arun Babu, Kushal Tirumala, Michihiro Yasunaga, Leonid Shamis, Jacob Kahn, Xuezhe Ma,
Luke Zettlemoyer, Omer Levy. arXiv:2408.11039, August 2024.
ALMA: Alignment with Minimal Annotation.
Michihiro Yasunaga, Leonid Shamis, Chunting Zhou, Andrew Cohen, Jason Weston, Luke Zettlemoyer, Marjan
Ghazvininejad. arXiv:2412.04305, December 2024.