Julien Delaunay

A dedicated researcher with a strong focus on Machine Learning, Natural Language Processing, and Human-Computer Interaction. I build production-grade LLM applications using LangChain, OpenAI, Vertex AI, and the Google Cloud / BigQuery ecosystem, specializing in advanced RAG systems, multi-agent orchestrators, and co-inventor of a patented medical AI scribe deployed across the Top Doctors Group.

AI Researcher at Top Health Tech Barcelona (2024 – Present)

• Lead development of LLM pipelines to convert unstructured PDF clinical reports into structured healthcare data.
• Built a custom multi-agent orchestrator using LangChain to query and interact with medical data.
• Designed a vocal AI agent orchestrator for medical appointments on the ElevenLabs Agent platform.
• Benchmarked and co-patented an ambient medical AI Scribe, deployed to all doctors across Top Doctors Group.

PhD Student at INRIA Rennes (2020 – 2023)

During my PhD on “Explainability for Machine Learning Models: From Data Adaptability to User Perception”, I designed and evaluated post-hoc explanation methods for multi-modal data. I developed algorithms to assess linear surrogate models, improving model trust and interpretability. Published 4+ papers in top-tier international venues (CSCW, CIKM, IDA) and collaborated across multiple European institutions.

Collaborative Researcher at Aalborg University (2022 – 2023)

I led international user studies to assess explainable AI (XAI) interfaces in high-stake domains like healthcare and law. Designed rigorous experimental protocols to analyze how different modalities (visual, text) and styles of explanation affect user understanding, cognitive load, and decision-making accuracy.

PhD Thesis – Rennes Univ. (2023)

Research Master – Rennes Univ. (2020)

Master in AI – Sherbrooke (2019)