Description du poste
募集の背景、プロジェクトの説明 [Outline of Laboratory / Division]
The RIKEN Center for Computational Science (R-CCS), led by Center Director Satoshi Matsuoka, are threefold, centered around supercomputing: one is to target high performance computation itself as a scientific objective, or namely, the "Science of Computing"; another is to apply the enormous computational power thus obtained to solve difficult scientific problems, or namely the "Science by Computing"; finally, to collaborate with other scientific disciplines that contribute to advances in both sciences, or namely, the "Science for Computing." Our goal is to be recognized as one of the global leadership research centers to advance high-end computational science in this regard.
The Materials Science Application Interface Platform Development Unit is recruiting research employees to carry out a new theme. The research overview of the laboratory is shown below. Our main target is as follows.
Development of Molecular Software and Theory through Foundation Models
1)Engineering of systems based on large language models to automatically generate quantum chemistry software components for use on FugakuNEXT. Creation of methods for automatic software documentation, validation, and optimization specifically for high-performance computing.
2)Use of multimodal models to generate hypotheses for new quantum chemistry methods or studies. Development of an Artificial Intelligence agent driven loop of hypotheses formulation, code generation, deployment, analysis, and visualization for molecular science.
[Keywords]: Agentic AI, electronic structure, materials science, quantum chemistry 仕事内容・職務内容 The recruited researcher will carry out a project supported by JST ARiSE projects. Their assigned work directly contributes to the Materials Science Application Interface Platform Development Unit’s research on integrating Large Language Models and Agentic AI into materials science research. The overall goal of the ARiSE project is to develop a foundation model for materials science. To this end, the creation of large datasets of high-quality materials simulations is essential. However, typical high throughput simulation work treats electronic structure methods as a black box, leading to datasets of uneven or even questionable quality. The primary purpose of this subtheme will be to develop agents able to automatically utilize multiple simulation codes and theories, while making the dynamic decisions required for high fidelity results. The researcher will also collaborate with members of the larger project towards the development of practical datasets and models. The researcher will further build and optimize the software components required for high-quality results to be generated efficiently on HPC platforms. External Fund: JST ARiSE “Development of Materials-AI Platform for "AI for Materials"”
The project funding period is 30 October 2026–31 March 2029. The job description may be changed during the contract period or upon contract renewal to the extent described above. 配属部署 Existing departments Materials Science Application Interface Platform Development Unit, AI for Science Platform Division, RIKEN Center for Computational Science 職種 Researcher/Postdoc level 研究分野 Life Science - Biophysics Natural Science - Biophysics, chemical physics and soft matter physics Informatics - High performance computing