Sepehr Akbari

me

Hi! I'm Sepehr a junior pursuing a bachelors in Computer Science and Mathematics at Lake Forest College. I'm driven by a mix of human-centered AI and creative problem-solving. While most of my coursework and projects lean toward deep learning and AI engineering, I've also worked on full-stack development (MERN), GPU computing with CUDA, and basic quantum computing using Qiskit. I'm fascinated by the "sweet spot" where algorithmic automation meets human innovation, building models that don't just process data, but expand our capacity for creativity.

My current research interest is in the intersection of deep learning and combinatorial optimization. I am particularly interested in leveraging diverse architectures, ranging from reinforcement learning to computer vision, as tools to develop neural heuristics for NP-hard problems in mathematics and computer science.

I thrive in the algorithmic world, but I'm just as interested in how technology connects with human innovation. I'm always eager to connect, learn, and collaborate on interesting problems in CS and Math, so please feel free to reach out!

Current Endeavours

Courses:

This semester my main focus is on my bachelor's thesis. which explores combining Transformers and Geometric Deep Learning to evaluate the computational complexity of generating Gröbner bases. My research involves learning a performance metric for Buchberger's algorithm, which is then integrated into a Proximal Policy Optimization (PPO) agent to develop an optimal generator selection strategy. While a niche area of study, I am excited by its potential to resolve computational bottlenecks in various fields, including efficient motion planning in robotics, faster geometric solving in computer vision, and robust risk modeling in quantitative finance.

To prepare myself for my thesis, I spent some time with IVA by Cox et al. last semester, which beyond just algebraic geometry, sparked a fascination with tensor geometry; leading me down the rabbit hole of studying Landsberg's Tensors this semester. I'm also diving into higher-level statistics this term right after studying probability theory; curious to see how these concepts bridge over to my previous work in Bayesian ML. To round things out, I'm taking an algorithms course, which hopefully adds some pure-CS time to my schedule, and a course dedicated to proving "Great Theorems" to satisfy my math tooth.

Additionally, I work as a part-time software engineer at the Applied Data Center, building internal tools for usage across various departments. I am also a teaching assistant for multiple CS/Math courses.