Battery AI & Digital Twin Engineer (Cell Design & Simulation)

Ampace (新能安) is a global battery-technology company specializing in lithium-ion cells and systems for energy storage, data centers, and critical power. Our R&D pairs advanced cell chemistry with semi-solid-state technology to deliver safe, high-performance batteries at scale. This position is based at our R&D center in Xiamen, China.

Responsibilities:

• Develop hybrid physics-ML models that predict and optimize cell energy density, cycle life, and thermal behavior.
• Map core design parameters — coating loading, calendering pressure, electrode porosity — to electrochemical fields such as liquid-phase diffusion, solid-phase overpotential, and heat flux.
• Extend electrochemical mechanistic models and fuse them with machine learning (physics-informed neural networks, neural operators) into fast multiphysics surrogate models.
• Lead inverse-design using active learning and Bayesian optimization to cut physical trial-and-error and the number of DOE runs.
• Design and maintain versioned, high-concurrency simulation data pipelines and an MLOps workflow.

Requirements:

• PhD in Computer Science, Applied Mathematics, Physics, Mechanical Engineering, or a related quantitative or computational field; strong Python and industrial-grade ML/DL project experience on physical or engineering data.
• Electrochemical mechanistic modeling (SPM, P2D/DFN) or thermal simulation; PyBaMM, COMSOL, or ANSYS a plus; deep understanding of energy-density calculation and aging mechanisms such as lithium plating and SEI growth.
• Mechanism-ML hybrid methods (PINNs, neural operators, surrogate solvers); Bayesian inference, uncertainty quantification, Bayesian optimization, and active learning; MLOps tooling (DVC, LakeFS, MLflow, W&B) a plus.

新能安(Ampace)是一家全球领先的电池科技企业,专注于锂离子电芯及电池系统的研发与制造,产品广泛应用于储能、数据中心及关键电源等领域。公司结合先进的电芯化学体系与半固态电池技术,致力于打造兼具高安全性、高性能和规模化生产能力的电池产品。

本岗位工作地点位于中国厦门研发中心。

岗位职责

• 开发融合物理机理与机器学习的混合模型,用于预测和优化电芯能量密度、循环寿命及热性能。
• 建立涂布载量、辊压压力、电极孔隙率等核心设计参数与液相扩散、固相过电位、热流密度等电化学场之间的映射关系。
• 基于电化学机理模型进行扩展开发,并结合机器学习方法(如物理信息神经网络(PINNs)、神经算子等),构建高效的多物理场代理模型。
• 基于主动学习(Active Learning)和贝叶斯优化(Bayesian Optimization)开展电池反向设计,减少实验试错成本及DOE(实验设计)次数,提高研发效率。
• 设计并维护支持版本管理、高并发计算的仿真数据平台,建立并持续优化MLOps开发与部署流程。

任职要求

• 计算机科学、应用数学、物理学、机械工程或相关计算类专业博士学历;熟练掌握Python,具备工业级机器学习或深度学习项目开发经验,有物理或工程数据建模经验。
• 熟悉电化学机理建模(如SPM、P2D/DFN模型)或热仿真,具有PyBaMM、COMSOL、ANSYS等工具使用经验者优先;深入理解电池能量密度计算方法及锂析出(Lithium Plating)、SEI膜生长等老化机理。
• 熟悉机理模型与机器学习融合方法,如物理信息神经网络(PINNs)、神经算子(Neural Operators)、代理模型(Surrogate Model)等;具备贝叶斯推断、不确定性量化、贝叶斯优化、主动学习等相关经验;熟悉DVC、LakeFS、MLflow、Weights & Biases(W&B)等MLOps工具者优先。