Clément Rouvroy | M.Sc. in Computer Science

Clément Rouvroy

Database & Reinforcement Learning @ UZH

I am a Computer Science master's student at ENS PSL in Paris, working at the intersection of Machine Learning and Data Systems. My main interests revolve around databases, optimization, and solving large-scale computation problems.

Currently, I am studying Adaptive Query Processing for Factorized Databases supervised by Prof. Dan Olteanu at the University of Zürich. Before this, I had the opportunity to research Column-Oriented Databases at Nanyang Technological University under the supervision of Prof. Gao Cong and Dr. Jiachen Shi. Before that, I learned a lot working on Query Answer Enumeration, with Nofar Carmeli, David Carral, and Luc Segoufin across the Valda Team at ENS-PSL and the Boreal Team at INRIA Montpellier.

Previously at

Education

Currently at

News

Currently none.

Research Interests

For a list of my talks and publications, please see my Research Page.

Database Systems

My research focuses on optimizing information retrieval and query execution in large-scale database systems. At UZH, I am currently developing an adaptive query processing engine enhanced by Reinforcement Learning techniques and Factorized Databases. Previously at NTU, I designed a What-If analysis tool for column-oriented databases. This system accurately estimates the performance benefits of adding a specific index to a database configuration without incurring the computational cost of physically building it and running the queries.

Database Theory

During my research at INRIA Montpellier and ENS-PSL, I investigated Constant-Delay Enumeration. The core challenge of this field can be summarized as follows: given a preprocessing phase on a database, is it possible to enumerate all answers to a given query with a strict O(1) delay between each result?

AI as a Tool

In my database systems research, I actively leverage Artificial Intelligence whenever it provides a tangible benefit—specifically, when the inference overhead is strictly outweighed by the performance gains. While this pragmatic approach often leads me to employ specialized tools beyond standard LLMs (such as regression models, MCTS, and RL), I maintain a strong foundation in Deep Learning. My broader AI experience encompasses Vision Models, Diffusion, GANs, LLMs, Deep-RL, and model Robustness.