ICLR 2026
AgentSynth: Scalable Task Generation for Generalist Computer-Use Agents
Generating long-horizon computer-use tasks by composing verified subtasks, with controllable difficulty for agent evaluation.
I am a quantitative researcher at Voleon. I completed my PhD at UC Berkeley, where I worked with Prof. Feng Wang on experimental condensed-matter physics.
My research spans machine learning and automated experimental systems, with recent work on AI agents, evaluation, and scaling laws. I am currently collaborating with Prof. Dawn Song on Agents’ Last Exam as a core contributor.
I am also exploring robot learning and sim-to-real transfer through hands-on manipulation experiments. I am interested in how learning systems generalize beyond their training data and interact with the physical world.
ICLR 2026
Generating long-horizon computer-use tasks by composing verified subtasks, with controllable difficulty for agent evaluation.
Research collaboration 2026
A benchmark of long-horizon professional tasks with verifiable outcomes, built to evaluate agents on real-world work.
arXiv 2026
Turning scientific codebases into executable environments with expert-defined checks for training and evaluating scientific agents.
Independent project 2026
A hands-on SO-101 manipulation study using ACT and π0.5 to explore how simulated demonstrations interact with limited real-world task coverage.
COLM LM4Sci workshop 2026
A repository-centered workflow for human–AI scientific prototyping, with evidence gates and auditable experiment records.
COLM Re-Data workshop 2026
Studying how repeated details can link identities across pseudonymized documents, and how corpus-aware generalization can reduce that risk.
COLM LSEI workshop 2026
Learning when a clarification question is worth asking by balancing useful information against interruption and error costs.
* Equal contribution
Nature 2026
Evidence for two-component exciton condensates in an electron–hole bilayer under magnetic fields.
Science 2025
Electrical control of interlayer three-particle bound states and exciton–trion mixtures in electron–hole bilayers.
Science 2025
Optical transport measurements reveal perfect Coulomb drag in a correlated electron–hole bilayer.
Science Advances 2025
An energy-efficient nonlinear optical device array brings neural-network activations to computing with incoherent light.
Nano Letters 2024
Charge-transfer doping at a van der Waals interface enables low-resistance contacts for atomically thin p-type transistors.
Nature 2024
Growing long, aligned graphene nanoribbons inside hexagonal boron nitride stacks for electronic devices.
Nature Materials 2024
A remote charge superlattice provides a tunable route to correlated electronic states in bilayer graphene.
Nature Materials 2021
Scanning tunneling microscopy connects atomic reconstruction to localized moiré flat-band states.
Nature Physics 2021
Real-space imaging resolves local charging and discharge cascades of correlated electrons in a moiré superlattice.