cv

AI research engineer

Valeo

Research engineer

VinAI

• Designed LiDAR-based 3D object detection models, achieving top-tier performance in the Waymo OD Challenge.
• Curated and utilized our own in-house large-scale VinScenes dataset and deployment on embedded platforms.

Researcher

Institut de Recherche en Informatique de Toulouse (IRIT, CNRS)

• Developed fusion-based 3D object detection algorithms integrating LiDAR and camera data, achieving robust performance in adverse weather conditions.
• Synthesized over 7k samples (LiDAR & images) within the MultiFogKITTI dataset by augmenting the existing KITTI dataset with synthetic fog conditions; this initiative addressed data imbalances and enhanced model accuracy for 3D object detection in adverse weather scenarios.
• Published and presented research findings at international conferences, journals.

Research engineer intern

Laboratoire de Vision et d'Apprentissage pour l'analyse de scène (LVA, CEA-List)

• Implemented and optimized YOLOv3 object detection model using PyTorch and TensorRT for high-performance Nvidia embedded platforms, doubling inference speed while keeping almost the same performance.
• Developed a robust object detection and recognition pipeline tailored for large-scale datasets, incorporating advanced depth-sensing sensor to enhance accuracy and reliability.

Research engineer intern

Laboratoire Mathématiques, Image et Applications (MIA)

• Reimplemented state-of-the-art 2D object detection models in TensorFlow (maiminh1996/YOLOv3-tensorflow (300 stars)) and fine-tuned them on custom boat datasets.
• Deployed the model for edge devices, Raspberry Pi 3 and TurtleBot 3.

PhD

Paul Sabatier University

Computer Science

Diplôme d'ingénieur (MSc)

INSA CVL

Electrical Engineering