NeurIPS 2026Spotlight
AI Researcher
Hi, I'm Do Hoang Lam.
I study how large models learn, reason, and fail — and build methods to make them more reliable, efficient, and aligned.
About
I'm an AI researcher working at the intersection of machine learning, natural language processing, and interpretability. My work asks why modern neural networks generalize, and how we can make their behavior more predictable.
Previously, I completed my degree at FPT University, and have interned at Some Lab. Outside of research I enjoy open-source, teaching, and long hikes.
Find me online
Research interests
LLM Reasoning
Chain-of-thought, planning, and test-time compute scaling.
Interpretability
Mechanistic analysis of circuits and representations in transformers.
Alignment & Safety
Robustness, evaluation, and scalable oversight of foundation models.
Efficient Learning
Data-efficient training, distillation, and sparse architectures.
News
- Oct 2026Paper accepted at NeurIPS 2026 as a spotlight.
- Jul 2026Started a research internship at Some Lab.
- Mar 2026Gave an invited talk at the ML Reading Group.
Selected publications
Projects
Let's collaborate
Open to research collaborations, talks, and discussions. The best way to reach me is by email.
dohoanglam2902@gmail.com