Experience

Laboratoire Informatique d'Avignon (LIA), Avignon University

• Research Services
• 1 - 100 Employee

• PhD Student in speaker verification and anti-spoofing at LIA

  • Sep 2023 - Present

  Avignon, Provence-Alpes-Côte d'Azur, France



EURECOM

• France
• Research Services
• 100 - 200 Employee

• Research Engineer

  • Oct 2021 - Sep 2022

  France - Developed graph-based machine learning models to detect fraud in bank transactions. - Researched solutions for unsupervised anomalous sound detection for machine condition monitoring. - Teaching Assistant for the AML course at EURECOM.



EURECOM

• France
• Research Services
• 100 - 200 Employee

• Research Engineer

  • Mar 2021 - Aug 2021

  France Thesis: Applied Graph Neural Networks (GNNs) for fraud transaction detection (funded by Oracle)



Viettel Research and Development Institute

• Software Engineer

  • Jun 2018 - Jun 2019

  Vietnam Researched and developed algorithms for Aircraft Tracking System of Radar - Aggregating information from different sensors, filtering noise, and providing continuous information of targets. - Estimate the trajectory of targets based on their historical track points. (Multiple-Target Tracking) Developed modules: - FIltering: Develop 3D non-linear Kalman Filters integrated with Interacting Multiple Model (IMM) for 3D tactic Radar. This new technique improve 50% of measurement… Show more Researched and developed algorithms for Aircraft Tracking System of Radar - Aggregating information from different sensors, filtering noise, and providing continuous information of targets. - Estimate the trajectory of targets based on their historical track points. (Multiple-Target Tracking) Developed modules: - FIltering: Develop 3D non-linear Kalman Filters integrated with Interacting Multiple Model (IMM) for 3D tactic Radar. This new technique improve 50% of measurement error of the filtering module. - Orbital Correlation: Matching target points belong to target's trajectories using the Global Nearest Neighbor algorithm. - Target-Noise classification: Detect whether an observation is a target's point or noise using ensemble deep neural networks (98% accuracy). - Simulation: Build a tool for simulating multiple aircrafts with different trajectories to evaluate tracking algorithms. Show less