experiences
Jun 2026 - Sep 2026
Redmond, WA, USA
Research Intern
Skills: Cybersecurity agents, proxy evaluation, benchmark design, LLM post-training, LoRA, reinforcement learning
May 2025 - August 2025
San Francisco, CA, USA
Research Engineer Intern
Skills: AI safety, alignment evaluation, LLM agents, reinforcement learning, representation engineering
Aug 2023 - Present
Nashville, TN, USA
Graduate Student Research Assistant
Vanderbilt University
- Hardening AI against security threats: developed defenses against adversarial/backdoor attacks in neural networks using formal logic verification and spatial localization in neural representations, reducing attack success from 100% to 0% (NDSS'25, ECCV'26).
- Advancing generalization and privacy in AI: designed a domain-generalization method to handle client-to-client domain shift without sacrificing privacy, and built a large-scale, sequential identity-unlearning pipeline for generative models, improving identity retention by 700× (ICDCS'25, WACV'26).
- Safeguarding LLMs and agents from injection and misuse: developing security benchmarks for prompt-injection and jailbreak attacks in tool-using, OS-interacting, and code agents, including INFUSE (40 tasks, 56 attack scenarios, 38,000+ attack trajectories across 19 model families); developing FLOUR for targeted knowledge unlearning, preserving 99.6% of retrain-oracle utility (EMNLP'26).
Skills: AI security, backdoor defenses, machine unlearning, LLM agents, prompt injection, benchmark design, federated learning
Sep 2022 - July 2023
Hanoi, Vietnam
Research Assistant
Vinuni-Illinois Smart Health Center
- Developed a novel imperceptible attack achieving 100% success rates with durable backdoor effects, bypassing defenses in federated learning.
- Conducted and published a comprehensive survey of backdoor attacks, defenses, and future directions in FL.
- Evaluated federated unlearning methods empirically using MNIST, CIFAR-10, and CIFAR-100, demonstrating challenges in balancing unlearn time and efficacy.
Skills: Federated learning, generative models, adversarial machine learning, backdoor attacks, machine unlearning, deep learning, predictive modeling
Oct 2022 - Dec 2022
Tokyo, Japan
AI Research Intern
National Institute of Advanced Industrial Science and Technology (AIST)
- Studied vulnerabilities in distributed training systems and built FedGrad, a penultimate-layer signal–based defense that reduces attack success by 90% and detects nearly all malicious clients, safeguarding FL systems even when up to 50% of clients are adversaries and data is highly non-IID.
Skills: Federated learning, adversarial machine learning, defense mechanisms, heterogeneous data, AI research
Dec 2021 - Aug 2022
Hanoi, Vietnam
Undergraduate Researcher
Vinuni-Illinois Smart Health Center
- Developed FedDRL, a deep reinforcement learning-based FL aggregator addressing data heterogeneity and cluster-skewed distributions, outperforming FedAvg and FedProx by up to 4.05% on various datasets.
- Designed the PIKA architecture, combining external prescription data and knowledge graphs to enhance pill image recognition, achieving up to a 34.1% F1-score improvement over baselines.
Skills: Federated learning, adversarial robustness, anomaly detection, deep reinforcement learning, graph modeling, object detection
May 2021 - Dec 2021
Hanoi, Vietnam
Software Engineer Intern
- Developed fast, reliable, customizable online storefronts for retail businesses (Front-end).
- Designed and reviewed technical architecture proposals for emerging features, including multilingual conversion and integrating databases to support multiple languages.
Skills: Front-end development, technical architecture design, multilingual systems, database integration, retail technology
Sep 2019 - Aug 2022
Hanoi, Vietnam
Undergraduate Researcher
- Designed a reinforcement learning-based optimization algorithm for wireless rechargeable sensor networks, optimizing charging locations and times to extend network lifetime by up to 8.29x over existing methods.
- Developed applications for Fi-Mi, an AI-driven mobile air quality monitoring and forecasting system, enabling real-time data processing and display from sensors mounted on buses in Vietnam.
Skills: Optimization, reinforcement learning, application programming, time-series forecasting and calibration
Jun 2020 - Feb 2021
Hanoi, Vietnam
Software Engineer Intern
MDC Software Solutions
- Developed a mobile-platform social network for product reviews, enhancing user engagement through seamless UI/UX design and providing real-time feedback features.
- Researched and deployed cutting-edge deep learning models (YoLo and StyleGAN) on iOS apps, enabling advanced features such as real-time image recognition and predictive analytics for enhanced user experience.
Skills: Mobile app development, UI/UX design, real-time systems, deep learning models, YoLo, StyleGAN, iOS development, predictive analytics