Samuel Reyd

PhD candidate at Télécom Paris

Portrait or research illustration

Contact

samuel.reyd@telecom-paris.fr

Institutions

Projects

See my GitHub profile

Research profile

I develop causal methods to understand and explain the behavior of complex AI systems.

My current work focuses on actual causation, causal explanations, and relevance in explanations, with applications to complex adaptive systems and machine learning systems.

I am currently looking for PhD internships and postdoctoral opportunities in fields related to causality, explainability, LLM agents, or world models.

Short bio

I am a PhD candidate at Télécom Paris (expected Nov. 2026), under the supervision of Ada Diaconescu and Jean-Louis Dessalles, working on causality and explainability in complex AI systems. My research focuses on identifying actual causes and generating meaningful explanations in adaptive systems. See a list of publications here.

I hold Master’s degrees from Télécom Paris and Polytechnique Montréal, where I worked on knowledge graphs and machine translation with pretrained language models.

Highlighted work

🧩

Relevant Causes for Causal Explanations in Complex Adaptive Systems

ACSOS 2025

A framework for identifying causally relevant events in complex adaptive systems, with a focus on explanation quality and interpretability.

paper code
🌐

Machine Translation using Knowledge Graphs and Pretrained Language Models

ISWC 2023

A contribution at the intersection of knowledge graphs, natural language processing, and pretrained language models.

paper code
⚙️

actualcauses

PyPI package · associated preprint under review · NeurIPS workshop

A Python package for actual cause identification, designed to support reproducible research on causal explanation and complex systems.

package preprint workshop code

Research directions

My PhD focuses on actual causation, causal explanation, and relevance-based explanation methods for complex adaptive systems and machine learning systems.
More broadly, I am interested in explainable AI, causal inference, machine learning, and neuro-symbolic approaches. During my PhD, I also explored topics such as causal abstraction, causal emergence, causal representation learning, and causal discovery. During my master's degree, I worked on neural machine translation, semantic web, knowledge graph embedding, and pretrained-language models.
My long-term objective is to contribute to AI and machine learning research, especially on topics related to causality, interpretability, and foundation models. I am particularly interested in bringing rigorous causal and explanatory methods closer to modern AI systems.