I lead the UniVerSE (Unification and Verification of configuration SpacEs) project, which aims to unify existing work on configuration spaces. The Universal Variability Language (UVL) was initially proposed as a standard language for defining the configuration spaces of software product lines, but it is now increasingly used to model variability in a broader range of systems. Despite this growing adoption, UVL remains fragmented: its semantics can vary depending on the context in which it is used and on the specific implementation. Moreover, UVL offers limited expressiveness for specifying constraints and does not fully leverage the extensive body of work developed by the constraint-programming and constraint-reasoning communities. With UniVerSE, I aim to strengthen UVL by extending its foundations and providing a unified formal semantics for the specification and verification of configuration spaces.
I am a co-leader of the RADyD (Resilience through Dynamic Adaptation for Data Processing) project. The purpose of the project is to investigate how data-processing systems can remain operational when their execution environment is degraded. The project explores semantics-aware resilience, where systems dynamically reconfigure their software components, execution strategies, or hardware resources while preserving the strongest possible semantic guarantees. In particular, RADyD studies how semantic contracts can characterize substitutions as exact, conditionally equivalent, or approximate with explicit quality bounds, turning adaptation into controlled and explainable degradation rather than ad hoc fallback.
I am a part of the ANR For-Coala project. The project aims to bring formal guarantees to Infrastructure-as-Code (IaC) languages used to configure and reconfigure distributed software systems. The project seeks to bridge the gap between widely adopted DevOps technologies and formally defined academic approaches, focusing in particular on Ansible and the academic language Concerto. Its goal is to develop mechanized formal semantics for these languages and verified, semantics-preserving transformations between them, ultimately providing open-source, formally certified configuration languages, execution engines, and translation tools.
As a researcher in the Taranis project, part of the PEPR Cloud initiative, I focus on rethinking the design and verification techniques associated with Infrastructure as Code (IaC) languages. The Taranis project aims to simplify the development and deployment of cloud applications. My work involves addressing the current gaps in tooling within the cloud community, particularly for languages like YAML, where there is a lack of advanced tools such as type systems, linters, and IDEs. This tooling is essential for making informed decisions regarding topology and resource allocation, considering various criteria such as DevOps (operational efficiency), SecDevOps (security), FinOps (financial optimization), and GreenOps (ecological impact).
As a postdoctoral researcher for the ANR SeMaFoR project, I contributed to WP3: Decentralized Reconfigurations, under the supervision of Hélène Coullon and Charles Prud'homme. The SeMaFoR project aims to make significant advances in the collaborative exploitation of Fog resources. I was responsible, for the task concerning my postdoctoral position, for the inference and planning of reconfiguration programs.
My thesis was funded by the Marie Skłodowska-Curie actions, through the Lowcomote project. Lowcomote is a European project aimed at training a generation of professionals in the design, development, and operation of low-code platforms. In this project, I participated in writing several deliverables, as well as in the overall coordination of the project by volunteering to represent the project's doctoral students and administering its website.
I was a research engineer in the Girafon project, funded by the Centre-Val de Loire Region (APR-IA), which aimed to propose algorithms for querying, exploiting, and analyzing large graphs using a rich query language and distributed computing. I contributed to the task "Structured parallel programming on large graphs".