Alexander Shmakov

Artificial Intelligence Researcher

Hi there! I am an AI researcher who loves everything deep learning and robotics. I specialize in applying generative AI to scientific challenges and leveraging reinforcement learning for planning. I've published papers on applications of deep learning to planning, robotic control, high energy physics, astronomy, chemical synthesis, and biology. Here are some fantastic organizations that I am part of who let me work on all of these projects.

University of California Irvine


Computer Science Ph.D.

Generative Transformer Models for Inverse Problems in Particle Physics

2020 - 2025 NSF MAPS Fellow 4.0 GPA

Computer Science B.S.

2015 - 2019 cum laude 3.92 GPA

Mathematics B.S.

2015 - 2019 magna cum laude Pi Mu Epsilon 3.88 GPA

Research Experience


UCI Institute for Genomics and Bioinformatics

Deep Learning Researcher
Jan 2018 - Present Lab Website

Conducting advanced AI research with applicability in particle physics, chemistry, astronomy, and biology. Inventing novel neural network architectures that incorporate physical symmetries and domain insights. Spearheading the development and execution of deep learning experiments, managing intricate cross-domain data sets, and efficiently employing large-scale compute clusters.

UCI Intelligent Dynamics Lab

Reinforcement Learning Researcher
Jun 2020 - Present Lab Website

Conducting in-depth research in deep reinforcement learning, targeting its practical applications in robotics, planning, and competitive multi-agent environments. Produced headlining results by engineering the first RL agent capable of learning to solve the Rubik's Cube without reliance on human knowledge.

ATLAS Collaboration at CERN

Visiting Researcher
Jan 2023 - Present Collaboration Website

Pioneering machine learning and generative AI innovations in high-energy physics to enhance detection of rare interactions involving the Higgs Boson. Integrating physical symmetries into state-of-the-art transformer and diffusion models. Developing software libraries widely employed across CERN to accelerate the search for new physics.

Amazon Science

Applied Science Intern
Jun 2024 - Dec 2024 Lab Website

Developed fully automated AI solutions to detect and adapt to fraud on a global scale. Designed innovative reinforcement learning techniques to identify shifts in customer behavior and swiftly adjust strategies to prevent losses. Led the implementation of interpretable fraud AI systems.

HPE Hewlett Packard Labs

AI Research Associate
Jun 2021 - May 2023 Lab Website

Pioneered the design of advanced multi-agent reinforcement learning systems for green energy solutions and industrial control mechanisms. Achieved breakthroughs in ensuring the safety and reliability of AI implementations, significantly reducing the cost of energy from wave-powered renewable energy sources.

Selected Publications


SPANet: Generalized Permutationless Set Assignment for Particle Physics
Shmakov A, Fenton M, Ho TW, Hsu SC, Whiteson D, Baldi P. (2022) • SciPost Physics
End-To-End Latent Variational Diffusion Models for Inverse Problems in High Energy Physics
Shmakov A, Greif K, Fenton M, Ghosh A, Baldi P, Whiteson D. (2023) • NeurIPS
Reconstruction of unstable heavy particles using deep symmetry-preserving attention networks
Fenton M, Shmakov A, Okawa H, et al. (2024) • Nature Communications Physics
Solving the Rubik's Cube with Deep Reinforcement Learning and Search
Agostinelli F, McAleer S, Shmakov A, Baldi P. (2019) • Nature Machine Intelligence
Multi-agent reinforcement learning controller to maximize energy efficiency for wave energy converter
Sarkar S, Gundecha V, Shmakov A, et al. (2023) • AAAI
AI for Interpretable Chemistry: Predicting Radical Mechanistic Pathways via Contrastive Learning
Tavakoli M, Baldi P, Carlton AM, Chiu Y, Shmakov A, Vranken DV. (2023) • NeurIPS