Tien-Dat Le
🇳🇱 EngD Researcher · Eindhoven University of Technology
Adversarially robust ML for safety-critical systems.
🇻🇳 Vietnam — security practitioner at Zalo (70M users, 5+ critical vulns) · 🇰🇷 Korea — built ML-based IDS with accuracy = 1.0 · 🇳🇱 Netherlands — now asking how to certify ML safety in production.
Research
My research sits at the intersection of time series anomaly detection, AI security / adversarial ML, and critical infrastructure security — specifically in-vehicle networks, power grids, and IoT systems.
I build models that are not just accurate, but certifiably robust when deployed under adversarial conditions. The through-line: offensive security intuitions applied to ML threat modeling, from a background of finding real vulnerabilities at scale.
Experience
- Developing AI-powered forecasting models for energy demand and renewable production prediction
- Integrating machine learning pipelines with operational energy management systems
- Benchmarking classical time-series models against deep learning approaches for short-term load forecasting
- Developing an Integrated Digital Platform for Multi-Energy Flexibility Asset Orchestration
- IoT-enabled asset integration combining edge computing with cloud analytics for real-time energy management
- Multi-objective optimization balancing technical, economic, and environmental constraints
- Digital twin architectures for simulating regional energy network scenarios
- GAN-based In-Vehicle IDS
- Multiclass classification using vision transformer + GAN (Auxiliary Classifier) for CAN bus anomaly detection
- Federated learning integration for privacy-preserving intrusion detection
- Target venue: IEEE TIFS (submitted 2026, under review)
- → [S.1]
- Multi-classification In-Vehicle IDS
- Transformer + autoencoder architecture for CAN bus traffic analysis
- Achieved classification accuracy = 1.0 on benchmark datasets
- → [J.1]
- AI-based Electricity Theft Detection
- Transformer + convolutional autoencoder for smart grid anomaly detection
- Achieved accuracy 0.9918 — state-of-the-art at time of publication
- → [J.2]
- Safety and Security Executive
- Conducted OWASP security audits of internal tools and external messaging platforms
- Identified and reported 5+ critical vulnerabilities affecting millions of users
- Safety & Security Fresher
- Vietnam's most-used messaging platform — 70+ million users
- Developed custom penetration testing platforms using JavaScript and Python (packet decryption, data visualization)
Publications
First Author
Contributing Author
Real-time power scheduling for an isolated microgrid with renewable energy and energy storage system via a supervised-learning-based strategy
Robust real-time energy management for a hydrogen refueling station using generative adversarial imitation learning
EfficientNet-based universum-inspired supervised contrastive learning and transfer learning for in-vehicle intrusion detection systems
Multi-class intrusion detection system for in-vehicle networks using few-shot learning and convolutional anomaly transformer network
Multi-Objective Energy Management for an Integrated Energy System With Small Modular Reactors Considering Uncertainty
Workshop / Conference
Vietnamese User Awareness Against Scams in Cyberspace: An Empirical Survey
Under Review
Federated Learning with Auxiliary Classifier GAN for In-Vehicle Intrusion Detection