Amirreza’s Page

Hi! I'm Amirreza Velae. I recently finished my B.Sc. in Electrical Engineering (with a minor in Applied Mathematics) at Sharif University of Technology, and I'm now an M.Sc. student in Electrical Engineering at KAIST. I'm fascinated by everything related to intelligence and robotics, especially reinforcement learning, optimization, and statistics. Outside of academics, you'll usually find me playing chess or soccer, or following chess tournaments. I'm also a big fan of movies and novels, though I don't get to enjoy them as much these days since I'm quite busy figuring out how an imaginary gambler should play against some fictional bandit machines. For a more formal introduction, please see the section below.

Feel free to reach out if you have a research opportunity, a basement full of spare GPUs, happen to be a time traveler, or just want to ask me about my favorite music band and share yours. I also love Persian rugs, so if you have one to show off, that would be great too!

Formal Bio

My name is Amirreza Velae. I hold a B.Sc. in Electrical Engineering, with a minor in Applied Mathematics, from Sharif University of Technology, and I am currently an M.Sc. student in Electrical Engineering at the Korea Advanced Institute of Science and Technology (KAIST), working with the U-AIM Lab under Prof. Chang D. Yoo. My academic interests center on the intersection of intelligence and computation, with a focus on modeling and realizing intelligence in machines. My primary research interest is reinforcement learning, which I view as a promising framework for advancing toward general intelligence. I am especially drawn to the theoretical foundations of deep reinforcement learning, statistics, and optimization.

For my B.Sc. thesis, I studied the numerical optimization behind Trust Region Policy Optimization (TRPO) under the supervision of Prof. Hamed Shah-Mansouri. Since April 2025, I have been leading a small research group on robust reinforcement learning under Prof. Sajjad Amini, conducting literature reviews and implementing algorithms. With Arash Bahari Kordabad and Prof. Sadegh Soudjani at the Max Planck Institute for Software Systems, I designed an algorithm for exact second-order updates in deterministic policy-gradient methods for the Linear Quadratic Regulator, published in Engineering Applications of Artificial Intelligence. I also collaborated with Prof. Mohammad Aliannejadi's group at the University of Amsterdam on debiasing retrieval systems, converting Backpack language models into encoders to mitigate gender bias in ranking; this work was accepted to ECIR 2026.

🎓 Education

• Korea Advanced Institute of Science and Technology (KAIST) - M.Sc. in Electrical Engineering, U-AIM Lab - Sep 2026 - Jun 2028 (expected)
• Sharif University of Technology - B.Sc. in Electrical Engineering, Minor in Applied Mathematics - Sep 2021 - Jun 2026
• Allameh Jafari High School (NODET) - Mathematics and Physics - 2018 - 2021

🔍 Research Interests

• Reinforcement Learning
• Statistics
• Optimization

💡 Selected Projects

• Controlling Gender Bias in Retrieval - with Prof. Mohammad Aliannejadi's group at the University of Amsterdam - ECIR 2026 / arXiv
• Second-Order Policy Gradient Method for the LQR - with Arash Bahari Kordabad & Prof. Sadegh Soudjani at the Max Planck Institute for Software Systems - Engineering Applications of AI / arXiv
• Numerical Optimization in Trust Region Policy Optimization (B.Sc. Thesis) - Advisor: Prof. Hamed Shah-Mansouri
• Robust Reinforcement Learning - Advisor: Prof. Sajjad Amini - Leading a small research group; literature review and algorithm implementation