Academic background
I have been a Postdoctoral Research Assistant at the Oxford Robotics Institute since August 2024. Previously, I was a Research Fellow in Machine Learning in the Autonomous Systems group at the Department of Computer Science, University College London (UCL).
My PhD in Computer Science, awarded in May 2023 from UCL, was sponsored by The Alan Turing Institute through its doctoral programme and included a research internship at Spotify. Prior to this, I attended the University of Birmingham between 2013 and 2018, obtaining an MSci in Computer Science with an Industrial Year.
Undergraduate Teaching
Statistics and probability, computer engineering, introductory programming.
Research Interests
In my research, I aim to develop artificial intelligence techniques for solving challenging decision-making problems effectively. This process generally involves taking a problem and formulating an appropriate mathematical model, creating an algorithm that learns solution strategies and becomes better with experience, and deploying it as an autonomous system capable of acting in the real world.
I am particularly fascinated by discrete problems on graphs, a powerful paradigm to reason about our connected world. More specifically, my research interests span reinforcement learning and planning, combinatorial optimization, network science, graph neural networks, game theory, and multi-agent systems. I also investigate applications of these techniques in areas as diverse as robotics, operations research, computer systems, and causal inference.
My main project at Oxford investigates mission planning techniques for underwater glider robots in collaboration with the National Oceanography Centre.
Hobbies
Outside of work, I like doing things that have nothing to do with artificial intelligence whatsoever. I am an aspiring polyglot and find language learning to be a very worthwhile activity for developing cross-cultural awareness and empathy. Other activities I enjoy are reading widely, listening to (live) music, expressing second-rate opinions about art, and lifting weights.
I am originally from the Moldavia region of Romania, where I completed my pre-university education. My Erdős–Bacon number is 8.
Publications
Selected Publications
- A Graph Reinforcement Learning Framework for Neural Adaptive Large Neighbourhood Search.SN Johnn, VA Darvariu, J Handl, J Kalcsics. Computers & Operations Research 172, 106791. 2024
- Reinforcement Learning Discovers Efficient Decentralized Graph Path Search Strategies.A Pisacane, VA Darvariu, M Musolesi. arXiv preprint arXiv:2409.07932. 2024
- Graph Reinforcement Learning for Combinatorial Optimization: A Survey and Unifying Perspective.VA Darvariu, S Hailes, M Musolesi. Transactions on Machine Learning Research (TMLR). 2024
- Large Language Models are Effective Priors for Causal Graph Discovery. VA Darvariu, S Hailes, M Musolesi. arXiv preprint arXiv:2405.13551. 2024
- PRORL: Proactive Resource Orchestrator for Open RANs Using Deep Reinforcement Learning. A Staffolani, VA Darvariu, L Foschini, M Girolami, P Bellavista, M Musolesi. IEEE Transactions on Network and Service Management. 2024
- Tree search in DAG space with model-based reinforcement learning for causal discovery VA Darvariu, S Hailes, M Musolesi. arXiv preprint arXiv:2310.13576. 2023
For an up-to-date publications list, please visit Victor-Alexandru’s Google Scholar Profile.
Links
Subject notes for courses taught at Jesus College:
See also Victor-Alexandru’s departmental profile and personal website.