mbmr.dev/
Hello there!
Hi! I am Mathis Randl, a PhD student in the SaCS lab at EPFL. I do research at the intersection of databases and machine learning, with a focus on improving the scalability and efficiency of large-scale inference.
These days, I'm (for the most part) working on software techniques for approximate vector search. Systems based on exotic hardware for vector retrieval are also in the works.
This website is meant to store my CV/resume, as well as archive papers, blog posts and announcements related to my professional works. Feel free to explore around!
Quick Links
Curriculum Vitae
Work in progress. Come back later!
Publications
(Authors typically listed alphabetically)
Proximity (Middleware '25)
Leveraging Approximate Caching for Retrieval-Augmented Generation
S. Bergman, A.-M. Kermarrec, D. Petrescu, R. Pires, M. Randl, M. de Vos, J. Zhang
[Download PDF] [Slides (WIP)][ArXiv] [ACM Digital Library (WIP)]
RagRoute (EuroMlSys '25)
Efficient Federated Search for Retrieval-Augmented Generation
R. Guerraoui, A.-M. Kermarrec, D. Petrescu, R. Pires, M. Randl, M. de Vos
[Download PDF] [Slides][Arxiv] [ACM Digital Library]
TeamFollowBack (ICWSM '24)
Detection & Analysis of Follow Back Accounts on Social Media
T. Elmas, M. Randl, Y. Attia
(A trip back to my dark data science era)
[Download PDF] [Arxiv] [AAAI library]