James Misaka (QuantumMisaka)

PhD researcher in AI4S & Computational Catalysis. I build ML potentials, atomistic simulation workflows, and open-source tools around the DeepModeling ecosystem.

Quick links

• Homepage: (In updating)
• GitHub: https://github.com/QuantumMisaka
• ORCID: https://orcid.org/0009-0007-8286-9090
• Email: quanmisaka [at] stu.pku.edu.cn

What I do

• Research: heterogeneous catalysis (iron-based Fischer–Tropsch), surface reactions, ML potentials (fine-tuning / active learning), atomistic simulations.
• Engineering: reproducible pipelines and tooling (e.g., DP‑EVA, transition-state workflows, ABACUS toolchain contributions).
• Community & leadership: open-source contributor/maintainer; community operator connecting developers and users; leading a small team building agentic tools for materials simulation.

Featured projects

Project	What it is
ATST-Tools — https://github.com/QuantumMisaka/ATST-Tools	Advanced ASE Transition State Tools for ABACUS and Deep Potential
DP‑EVA (dpeva) — https://github.com/QuantumMisaka/dpeva	Deep Potential Evolution Accelerator (workflow / active learning / representative data selection)
ABACUS user guide — https://github.com/MCresearch/abacus-user-guide	Chinese documentation & onboarding for ABACUS
ABACUS toolchain (upstream) — https://github.com/deepmodeling/abacus-develop/tree/develop/tools/toolchain	Cross-platform installation automation (major contributor)

Now (current quest)

• Shipping research-grade workflows that feel like production software: opinionated defaults, reproducibility, and good UX for scientists.
• Building agentic tools for materials simulation (planning, verification, and automation).