Agent architecture
Stateful multi-agent workflows, tool routing, specialised solvers, verification, and iterative refinement.
Maël JullienAI R&D Engineer
Agentic systems for complex NLP tasks.
I build and evaluate agentic AI systems, retrieval-augmented reasoning methods, and dependable frameworks for tasks that require expert-level language understanding.
Interested in advanced AI engineering and research collaborations in agentic systems, evaluation, and grounded NLP.
About
I build AI systems for complex language tasks where performance depends on coordinating retrieval, reasoning, tools, and verification.
My PhD at the University of Manchester used clinical trial language as a demanding testbed for evidence retrieval and inference. The public project archive documents that academic work: datasets, shared tasks, controlled model studies, retrieval methods, and specialised reasoning agents.
Capabilities & background
What I build
Grounded reasoning
Retrieval pipelines, evidence ranking, source attribution, and explicit inference over complex documents.
Evaluation systems
Reproducible benchmarks, objective scoring, diagnostic datasets, regression analysis, and model comparison.
Experience & education
Now
Luminance
R&D Engineer
Agentic AI systems, evaluation, and language technology
2026
Idiap Research Institute
Research Intern: Doctorate
Agentic reasoning frameworks
2026
University of Manchester
PhD Computer Science
Clinical NLI, retrieval, and controlled reasoning
2020
University of Nottingham
MSc Computer Science & AI
Distinction
2019
University of Leicester
BSc Mathematics
First Class Honours
Academic research
Clinical NLP as a testbed for reliable reasoning.
The archive contains my thesis and papers on clinical inference, benchmark design, retrieval, model evaluation, and structured agentic methods.
View publications