Applied AI for knowledge-intensive domains

Ngoc Luyen Le

I design ontology-grounded and agentic AI systems that combine LLMs, knowledge graphs, and explainable recommendation for skill intelligence and personalized learning.

AI Engineering LLMs Agentic AI Knowledge Graphs
Current focus LLMs, agentic AI, grounded skill search, competency modeling, and personalized training recommendation.
37 listed publications
10+ years in research and engineering
PhD Computer Science, IMT Atlantique
IEEE Senior Member, 2026

What I Work On

My recent work focuses on grounded AI systems for education, skill management, vehicle recommendation, and semantic search. I am interested in systems that can explain their recommendations, respect domain knowledge, and adapt to user context.

Research Path

I previously held a postdoctoral position at Heudiasyc, Université de Technologie de Compiègne. I earned my PhD at IMT Atlantique, where my doctoral research explored semantic representation for French with CCG and DRT.

Research Agenda

Ontology-grounded LLM systems

Combining domain ontologies, knowledge bases, and language models for grounded reasoning and explanation.

Agentic AI for education and skill intelligence

Tool-using AI workflows for skill intelligence, learner guidance, and human-in-the-loop decision support.

Knowledge graphs for recommendation and explanation

Semantic representations that make recommendation, search, and personalization more transparent.

AI-assisted learning, training, and competency modeling

Competency modeling, training recommendation, and learning trace analysis for personalized progression.

Selected Case Studies

View all projects
Skiller

Skill management with KG + LLM fallback

Competency modeling, grounded skill search, explainable recommendations, and personalized training recommendation.

EmONat

Graph-based tasks and lightweight memory

A web-first task workspace where each task keeps context, connected notes, and a small memory layer.

Vivocaz

Semantic vehicle recommendation

Ontologies, RDF similarity, user profiles, and context-aware recommendation for used-vehicle search.

Recent Research Outputs

View all publications