Scaling Frontier Physics Research
Quantum Materials Research
Autonomous Experiment System
Quantum Materials Research
Quantum Materials Research
My Publications
Autonomous Experiment System
Autonomous Experiment System
Unmodeled states and uncertain action outcomes in agentic scanning tunneling microscopy
Frontier LLM agents conditioned STM tips end to end on a real microscope. The task has traditionally been hands-on expert work. After a short apprenticeship, the agents worked alone and passed independent verification.
Evaluating Verified Autonomy in Quantum Engineering
Quantum-Harbor is a quantum lab where AI agents run experiments on simulated quantum hardware. A hidden grader checks every answer independently and never takes the agent's word for it. QIQCBench uses the lab to test 17 frontier agentic systems on 49 expert-authored tasks.
Abstraction Drivers for Scientific Instruments
quail-cli-core is a kit for building abstraction-layer drivers for scientific instruments. Its design is aligned with the Model Hardware Standard (MHS). All drivers share five commands, so an agent that has used one driver already knows how to command any other.
Explorations
Explorations
BGS Modding Superpowers
A plugin for coding harnesses such as OpenCode, Claude Code, and Codex. It lets AI agents work hands-on in mod development for Bethesda games.
LLM-Driven Open-World Simulation Engine
Built a state-management harness that lets a stateless LLM run a consistent evolving open-world game across long sessions.