About


Hi, I'm Duy! I'm a third-year PhD student in Computer Science at the University of Illinois Urbana-Champaign, advised by Prof. Minh N. Do and co-advised by Prof. Khoa D. Doan (VinUniversity). My research asks how to make multimodal and generative models dependable outside curated benchmarks: when a modality goes missing, when inputs are corrupted, when labels are scarce, or when a scene holds many people at once.

My path into AI started in Hanoi, where I was born and raised. As an undergraduate at Hanoi University of Science and Technology, I spent three years at the BKAI lab forecasting river levels and air quality from cheap, noisy sensors. Along the way I won a hackathon with a flood-forecasting platform. At the VinUni-Illinois Smart Health Center I built a pill-detection system that uses prescriptions as extra context. An internship at AIST in Japan, on multimodal learning with missing data, set the question I brought with me to Illinois.

Today I work end to end. I design methods (contrastive and self-supervised objectives, uncertainty estimation, diffusion distillation, RL fine-tuning) and evaluations that measure more than a single headline score. I also handle the engineering to train and evaluate large models on GPU clusters. My papers have appeared at IJCAI, ACL Findings and ICDM, with ACCV 2026 to come. Since 2025 I've been a Research Scientist Intern at Cantina Labs, working on image and video diffusion models.

I'm looking for Summer 2027 research internships in multimodal learning, generative models and trustworthy AI. Email is the fastest way to reach me, and I'm always up for a research chat.

Education


2023 – 2028 (expected) · University of Illinois Urbana-Champaign

PhD, Computer Science

Advisor: Prof. Minh N. Do · Co-advisor: Prof. Khoa D. Doan (VinUniversity) · GPA 3.98 / 4.00

2018 – 2022 · Hanoi University of Science and Technology

Engineer's Degree, Information and Communication Technology

GPA 3.77 / 4.00, top 3.3% of the School of ICT. Thesis: a knowledge-graph-assisted framework for pill detection (grade 4.00 / 4.00).

Experience


May 2025 – now · Cantina Labs (remote)

Research Scientist Intern

Distilling in-house image and video diffusion models for faster inference while preserving quality. Identity-preserving image generation, and audio-driven lip-sync video generation for human and non-human characters.

Jan 2021 – Aug 2023 · VinUni-Illinois Smart Health Center

Research Assistant, VAIPE project

Knowledge-graph-assisted pill detection combining graph neural networks with object detectors. Co-organized the AI4VN VAIPE Challenge 2023 (dataset, metrics, automated scorer).

Oct – Dec 2022 · AIST, Japan

Research Intern

Multimodal learning with missing modalities for healthcare applications.

Sep 2019 – Sep 2022 · BKAI Research Center, HUST

Research Assistant

Time-series forecasting for hydrology. Led the back-end and mobile teams of Fi-Mi, an AI-based air-quality monitoring and forecasting system.

Service


Program committee · Conference reviewer

NeurIPS, ICML, ICLR (2023 – 2026) · CVPR (2024 – 2026) · AAAI (2023, 2026) · KDD 2024

Honours


2022 · Vietnam

Second Prize, National Science and Technology Awards for Students in Higher Education Institutions

2022 · Vietnam AI Day

Top 5 AI Solutions, AI Awards 2022

VAIPE: a healthcare monitoring and support system.

2022 · Hanoi University of Science and Technology

Best Presentation Award, Undergraduate Thesis Defense

2021 · SoICT-IBM Hackathon

First Prize, Call for Code Track

A flood-forecasting platform built on sensor networks and deep neural networks.