Reza Sourki

Siemens, x UBC, Harvard

Hello and Welcome!

I’m interested in the interface of AI and physical simulations, and developing computational and data-driven solutions. I use the state-of-the-art AI algorithms and tools to learn and understand the physical systems. I utilize a range of techniques from mathematical models to AI to exploit the physical informations.

My expertise in applied mechanics, mathematics, and AI enables me to collaborate with various groups. I have been lucky to work on several ongoing multidisciplinary projects and research areas, which has broaden my range of expertise.

my research

Slide 1
Slide 2
Slide 3
Slide 4
previous arrowprevious arrow
next arrownext arrow

Currently, I’m focused on the state-of-the-art computation and physics AI. Previously, I obtained my Ph.D. from The University of British Columbia, and I did part of my Ph.D. at the School of Engineering and Applied Sciences at Harvard University.
In my Ph.D., I worked on soft, textile-based materials and assistive devices in Composites Research Network and Bertoldi Group advised by Drs. Katia Bertoldi, Goran Fernlund (UBC Emiratus Prof.), Keith Humfeld (Boeing), Abbas Milani, and Reza Vaziri.

My Ph.D. research aimed to develop theoretical and numerical methods to design and optimize materials. I have developed models that enable an accurate and efficient prediction of the behavior and the 3D shape of complex architected materials.
To do so, I have designed several advanced experimental setups and fixtures to capture the behavior of such materials under in-plane and out-of-plane deformation modes, and minimized the defect formations, i.e., wrinkles and bridges during deformation of soft, textile-based materials.
I used FE simulations and machine learning-based analysis, in particular, optimization algorithms, Reinforcement Learning, and Deep Learning to tackle problems.

my projects

GNN and physical domain

View →

PINNs

View →

recent posts

  • Assistive devices

    Most developments on wearables have been focused on rehabilitation devices [1,2], addressing effects of impairments that can potentially be reduced with time and training. Another branch of developments focused on assistive devices for individuals where full recovery is not likely, e.g., stroke survivors after reaching the recovery plateau [3] or…

  • Physics-Informed Neural Networks: Intro

    Code for a few simple examples is available in a colab notebook on my GitHub, and may be updated over time.  The Evolution of Physics-Informed Learning: Literature Review Deep learning has impacted many fields with its ability to find patterns in massive datasets. However, when it comes to modeling complex…

  • How I Learn (and How You Might, Too)

    Over the years, many people have asked me about my approach to learning. After more than two decades of continuous learning and over a decade of teaching experience, I paused to take a closer look at the strategies I rely on to learn efficiently. In another post, I shared my…

  • Projects to real-world change

    In the past decade, I’ve been trying to build, create, and improve. Here are a few takeaway messages I’ve learned for advancing projects. Compounding effect Focus on projects that genuinely interest you and that you enjoy. While it can be effective to aim for a specific goal with your projects,…