Research Fellow (Physics-Informed AI and Computational Materials Mechanics)
Nanyang Technological University · Singapore
About this role
The School of Mechanical & Aerospace Engineering (MAE) at Nanyang Technological University (NTU) is seeking a Research Fellow to develop physics-informed AI models for the discovery of damage-tolerant structural alloys for extreme environments. This role focuses on integrating AI, computational mechanics, and experimental data to predict deformation and fracture behavior, ultimately guiding microstructure and additive manufacturing process optimization. Key Responsibilities: Develop physics-informed AI and surrogate models for predicting deformation and fracture in structural alloys. Integrate AI with crystal plasticity finite element and phase-field simulations. Combine simulation and experimental data for model training, calibration, and validation. Apply inverse design and optimization to identify damage-tolerant microstructures and guide manufacturing processes. Collaborate with experimental researchers to interpret mechanisms and validate model predictions. Assess model accuracy, uncertainty, and applicability across microstructures and environmental conditions. Publish research findings, contribute to project reports, and mentor students. Required Qualifications: PhD in Materials Science and Engineering, Mechanical Engineering, Computational Mechanics, or a closely related field. Strong knowledge of computational mechanics or computational materials science, with hands-on experience in crystal plasticity finite element modeling and/or phase-field simulations. Experience in developing AI or machine learning models for physics-based scientific or engineering applications. Proficiency in Python, scientific computing, and machine learning frameworks such as PyTorch or TensorFlow. Good understanding of microstructure–property relationships and deformation mechanisms in structural materials. Preferred Qualifications: Knowledge of fracture mechanics is highly desirable. Experience with simulation–experiment data integration, physics-informed learning, inverse design, or uncertainty quantification would be advantageous. Strong analytical and communication skills, with the ability to work independently and collaborate across disciplines.
For full details about this role, click “Apply on company site” to view the employer’s original posting.
A focused preparation guide, built around this job’s responsibilities and requirements.
✦ AI-generated guide
Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.
01
Understand the role
What matters most
Development of Physics-Informed AI Models
The core objective is to build surrogate models that predict deformation and fracture in structural alloys. You must demonstrate the ability to embed physical laws into machine learning architectures, moving beyond black-box models to ensure predictions are physically consistent with material behavior.
Integration of Multi-Source Data
You will be responsible for merging high-fidelity simulation data (crystal plasticity finite element or phase-field) with experimental datasets. Success depends on your ability to calibrate and validate models using disparate data types to ensure accuracy across various microstructures.
Inverse Design and Optimization
The role requires applying optimization techniques to identify damage-tolerant microstructures. You must be prepared to use your models to guide additive manufacturing processes, effectively closing the loop between computational design and physical fabrication.
Likely challenges to think through
Bridging Computational Mechanics and AI
A significant challenge is the technical translation between traditional finite element/phase-field simulations and modern AI frameworks like PyTorch or TensorFlow. Prepare by being ready to discuss how you handle the computational cost of high-fidelity simulations while training efficient surrogate models.
Cross-Disciplinary Collaboration
You must collaborate with experimental researchers to validate your models. The difficulty lies in interpreting experimental noise and translating physical observations into actionable constraints for your AI models. Prepare by highlighting examples of how you have communicated complex computational findings to non-computational peers.
Uncertainty Quantification
Assessing model applicability across diverse environmental conditions and microstructures is a key requirement. You will face the challenge of quantifying uncertainty in predictions where experimental data may be sparse. Prepare by reviewing methods for robust uncertainty estimation in physics-informed neural networks.
03
Practice your answers
Technical practice
Integrating Physics-Informed AI with Crystal Plasticity
What to coverExplain how you would incorporate physical constraints, such as crystal plasticity constitutive equations, into a neural network loss function. Discuss the trade-offs between purely data-driven models and physics-informed architectures in terms of data efficiency and generalization.
Surrogate Modeling for Fracture Mechanics
What to coverDescribe your approach to developing surrogate models for predicting fracture behavior in structural alloys. Address how you handle the high dimensionality of microstructure data and the computational cost of generating training data from phase-field simulations.
Simulation and Experimental Data Fusion
What to coverHow do you approach the calibration and validation of a computational model when simulation results and experimental data exhibit discrepancies? Discuss techniques for data integration and how you quantify the reliability of the resulting model.
Inverse Design for Microstructure Optimization
What to coverExplain the methodology you would use to perform inverse design to identify damage-tolerant microstructures. Discuss the challenges of navigating a large design space and how you ensure the optimized microstructures are manufacturable via additive processes.
Uncertainty Quantification in Materials Modeling
What to coverWhy is uncertainty quantification critical when predicting deformation in extreme environments? Suggest methods for assessing model accuracy and uncertainty across varying microstructural and environmental conditions.
Python Frameworks for Scientific Computing
What to coverCompare the strengths of PyTorch versus TensorFlow for implementing custom physics-informed loss functions. Provide examples of how you have utilized scientific computing libraries in Python to handle complex material mechanics datasets.
Microstructure-Property Relationships
What to coverDiscuss your understanding of how specific microstructural features influence the fracture toughness of structural alloys. Explain how you would represent these features as inputs for a machine learning model.
Phase-Field Simulation Fundamentals
What to coverDescribe the role of phase-field modeling in predicting crack propagation. How do you ensure the stability and convergence of these simulations when coupled with AI-driven optimization loops?
Behavioural practice
Cross-Disciplinary Collaboration
What to coverDescribe a situation where you had to collaborate with experimental researchers to validate a computational model. How did you reconcile differences in terminology or methodology to ensure the model accurately reflected physical reality?
Mentorship and Knowledge Sharing
What to coverProvide an example of a time you mentored a student or junior researcher on a complex technical project. How did you break down advanced concepts like physics-informed AI to ensure their understanding and contribution?
Independent Research Problem Solving
What to coverTell me about a time you encountered a significant roadblock in your research, such as a model failing to converge or data being insufficient. What steps did you take to diagnose the issue and pivot your strategy independently?
Communicating Complex Findings
What to coverDescribe a time you had to present complex computational findings to an audience with varying levels of technical expertise. How did you tailor your communication to ensure the impact of your research was clearly understood?
02
Sharpen your knowledge
Topics to prepare
Physics-Informed Neural Networks (PINNs)
Review the integration of physical laws into loss functions for neural networks. Exercise: Implement a simple PINN to solve a 1D heat equation or a basic mechanics problem to demonstrate how constraints improve model physical consistency, which is central to the role's requirement for predicting deformation and fracture.
Microstructure-Property Relationships in Structural Alloys
Refresh knowledge on how grain structure and phase distribution influence mechanical performance. Exercise: Summarize the impact of specific microstructural features on damage tolerance in extreme environments, preparing you to discuss how your models will guide additive manufacturing process optimization.
Uncertainty Quantification (UQ) in Scientific Machine Learning
Study methods for assessing model reliability, such as Bayesian neural networks or ensemble techniques. Exercise: Outline a strategy for quantifying prediction uncertainty in a surrogate model, directly addressing the requirement to assess model applicability across varying environmental conditions.
Skills to revise
Crystal Plasticity Finite Element Method (CPFEM)
Practice setting up and running CPFEM simulations to model material deformation. Exercise: Run a simulation of a representative volume element (RVE) under uniaxial tension to extract stress-strain data, which is essential for the required integration of simulation data with AI training pipelines.
Inverse Design and Optimization
Develop proficiency in using optimization algorithms to identify material parameters that meet target performance criteria. Exercise: Use a Python-based optimization library to perform a parameter sweep or inverse search for a microstructure that minimizes a specific fracture metric, aligning with the role's focus on discovering damage-tolerant alloys.
PyTorch/TensorFlow for Physics-Based Applications
Build custom data loaders and loss functions tailored for scientific datasets. Exercise: Create a training script that incorporates a physical constraint (e.g., mass conservation or boundary conditions) into the loss function, demonstrating your ability to bridge the gap between standard machine learning and computational mechanics.
04
Your preparation checklist
Synthesize your computational mechanics and AI experience
Prepare a concise summary of your PhD research that highlights the intersection of crystal plasticity finite element (CPFE) or phase-field simulations with machine learning. Be ready to explain how you have integrated these methods to solve specific materials science problems, such as predicting deformation or fracture behavior.
Prepare a technical project walkthrough
Select a past project where you developed a physics-informed AI or surrogate model. Structure your presentation to cover the data source (simulation vs. experimental), the choice of framework (PyTorch or TensorFlow), the physics constraints applied, and how you validated the model's accuracy and uncertainty.
Demonstrate inverse design and optimization proficiency
Review your experience with inverse design techniques. Prepare to discuss how you have used optimization algorithms to identify optimal microstructures or guide manufacturing processes, specifically focusing on how you balance computational efficiency with physical accuracy.
Formulate questions on interdisciplinary collaboration
Prepare thoughtful questions for the interviewers regarding the current workflow between the computational team and experimental researchers at NTU. Ask about the specific challenges in integrating experimental data with simulation-based training sets for structural alloys.
Review microstructure-property relationships
Refresh your knowledge on the deformation mechanisms of structural alloys in extreme environments. Be prepared to discuss how your computational models account for these physical phenomena and how you handle the transition from micro-scale simulations to macro-scale material performance.
Availability last checked: 18 Oct 2026. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy
Share this opportunityKnow someone who would be a great fit?