Aleksandar Todorov

About

I am a Master's student in Artificial Intelligence and Applied Mathematics at the University of Groningen, where I also completed my Bachelor's degree in Artificial Intelligence with the highest distinction Summa Cum Laude and graduated first in my cohort. I am motivated by the growing gap between what modern learning systems can do and our conceptual and mathematical understanding of why they work. I am interested in questions that try to close this gap and where understanding the structure of learning systems can make them more efficient, reliable, and amenable to analysis.

Beyond research and work, I try to maintain a balanced and healthy lifestyle, both physically and mentally. My cat, Beans (who you might spot on my GitHub profile picture), i.e., my unofficial co-researcher and mental health supervisor, plays an important role in keeping me grounded and reminding me to take breaks by ensuring I cannot hyperfocus on one thing for too long.

Research Interests

My interests lie primarily in continual learning, optimization, reinforcement learning and control. More broadly, I am interested in how learning systems can continue adapting as their experience changes while operating under constraints on memory and computation, preserving what remains useful, and allowing us to understand when and why their behavior may become unreliable. In terms of methodology, I like drawing on ideas from convex and online optimization and systems and control theory. In the past few years, I have been enjoying JAX more and more, especially for reinforcement learning.

I am currently working on projects in continual and reinforcement learning, and am looking for a PhD position, ideally starting spring, summer, or autumn 2027.

Current Work

• Research Operations - Safe AI Netherlands
  Starting September 2026 · the Netherlands
  Starting September 2026, I will be joining Safe AI Netherlands as a Research Operations, where I will be coordinating the national research hub by connecting researchers with collaborators, tracking projects, and arranging access to compute and funding.
• Visiting Research Scholar - KU Leuven
  July 2026 - August 2026 · Leuven, Belgium
  During the summer of 2026, I had the opportunity to work as a visiting research scholar at KU Leuven, where I collaborated on a project related to continual learning through more efficient optimization.
• AI Engineer - DataNorth
  March 2025 - present · Groningen, the Netherlands
  I mostly try to build and automate stuff in a scalable way, ranging from custom document extraction models to computer vision solutions for manufacturing. My work spans the full stack, integrating Azure (including Azure Document Intelligence), LangChain, Terraform, the TypeScript/Python ecosystem (Next.js, Docker, Tailwind, FastAPI), and many other frameworks.
• Teaching Assistant - University of Groningen
  September 2023 - present · Groningen, the Netherlands
  Teaching has been something that I have always enjoyed, and I have been fortunate enough to teach more than 15 courses at the University of Groningen so far, both at a Bachelor's and Master's level and for both the Artificial Intelligence and Applied Mathematics programs, in the fields of optimization, machine and reinforcement learning, logic, AI engineering, signal processing, linguistics, and natural language processing.
  For the academic year 2024-2025, I was also awarded the Teaching Assistant of the Year.