Quantum materials × Topology × AI for science.
I study exotic phases in low-dimensional systems (FCIs/FQAH, topological superconductivity) and build agentic AI tools that accelerate both experiment and theory.
- Exotic topological quantum matter aimed at computation-changing platforms (anyonics, strongly correlated phases).
- AI for labs: agentic systems that read instruments, control microscopes/stages, triage literature, and run analyses end-to-end.
For this mission, AI is necessary—to explore vast hypothesis spaces, automate measurement/analysis loops, and close the gap between theory and experiment.
labAgent— an agentic lab-management system for research automation (arXiv triage, tool-use, workflows, CI for science). https://github.com/caidish/labAgentinstrMCP— an MCP server for physics instrumentation that lets LLMs connect to (and safely interact with) quantum-device measurement setups. https://github.com/caidish/instrMCPtwoDClassifier— an MCP tool that distinguishes 2D materials, a necessary building block for my quantum devices. https://github.com/caidish/twoDClassifier
The current bottleneck is reasoning. Today’s models are not yet “smart enough” for fully autonomous scientific discovery.
I’m pushing toward super-AGI-level scientific agents via:
- high-quality data generation,
- robust tool-use policies and safety rails,
- rigorous evaluations and private benchmarks for scientific tasks.
- AI-for-Science projects (agentic tool-use, planning, evals, safety).
- Data generation that actually improves reasoning toward super-AGI.
- Quantum matters (2D/moiré, FCIs/FQAH, topological superconductivity).
- Instrumentation that microscopes quantum matters and AI at MACRO-, Meso, and nano-scales.
Physics •
AI agents and MCP servers for lab hardware •
Symbolic derivation at scale •
Building datasets/evals for scientific LLMs/agents.
I build everything from hardware (so HDL) to software (so frontend) even unusual Xware (e.g., symbolic tools), and I am willing to learn everything.
Email: jiaqic [at] mit.edu
If you’re exploring topological matter or building agentic tooling for labs, let’s talk. I’m happy to co-design evaluations, data pipelines, or instrument integrations.