At Sigservice, I developed and maintained a production product-recognition pipeline, from video processing to embedding generation and retrieval, using YOLO, SigLIP, Qdrant, Airflow, and GCP. I also developed recommendation and sales forecasting models.
At LISSI Laboratory of UPEC, I developed and evaluated deep learning and representation learning models for multimodal AI research. I prepared complex datasets and developed end-to-end AI solutions combining vision, language, and speech through LLM/RAG pipelines.
My research projects include a PDF-based Agentic RAG system with document ingestion, retrieval, RAG evaluation, and prompt-injection defenses, as well as a real-time conversational AI pipeline for speech interaction.
I’m the first author of an IEEE CASE publication on an incremental learning framework for plant identification in airports. I’ve also taught Deep Learning practical sessions and Introduction to AI.
