Hello!
[Github] [Twitter] [LinkedIn] [Google Scholar]
I am a third year CS PhD student at University of California, San Diego where I am advised by Prof. Lianhui Qin.
I completed my undergraduate studies in Computer Science at University of California, Santa Barbara where I worked with Prof. William Wang and the UCSB NLP group. me lol
Research Interests
My research focuses on building AI systems that keep getting better on their own. I am particularly interested in self-improving agents: harness evolution, where agents refine their own scaffolds, tools, and context; recursive self-improvement; and continual learning methods that let models acquire new skills and reusable abstractions on the fly. I am also excited about applying these systems to autoresearch and AI for science, with the ultimate goal of building agents that work alongside human experts on frontier problems in math, science, and engineering.
Publications
Ho et al. Agentic Context Management for Multi-Turn Human-AI Conversations. Preprint.
- paper | code (coming soon)
- Work done in collaboration with Microsoft Research.
Ho, Liu, Chen, Wang, Qin. Auto-Configuring Scientific Simulators with Lightweight Coding-Agent Adapters. NeurIPS 2026.
- paper
- Supported by the UC Geophysics x AI grant.
Ho et al. ArcMemo: Abstract Reasoning Composition with Lifelong LLM Memory. Preprint. ARC Prize 2025 Paper Track Runner Up
- paper | code | tweet | competition
Ho et al. Proof Flow: Preliminary Study on Generative Flow Network Language Model Tuning for Formal Reasoning. NeurIPS 2024 System 2 Reasoning at Scale Workshop.
Ho et al. WikiWhy: Answering and Explaining Cause-and-Effect Questions. ICLR 2023, Oral (Top 1.2% of 4019 submissions)