About This Role — The Mission
Precision product development is one of the last frontiers where AI has not yet been systematically deployed at the physics level. As a Technologist on this team, you will be one of the core builders who make physics-informed AI work in product development: not as a research prototype, but as models that engineers rely on to make real decisions. You will own key modelling workstreams end to end, from problem definition with domain scientists through to deployment. If you want to be hands-on with hard scientific ML problems and see your models drive real engineering outcomes, we want to talk to you.
Key Responsibilities
Deep Learning Model Ownership:
Design, build and own CNN, U-Net and ViT models for precision inspection, measurement and defect classification. Own model performance end to end: ablation studies, confidence calibration, and monitoring once deployed.
Anomaly Detection on Sensor and Process Data:
Build and own real-time anomaly detection on sensor, time-series and process data. That covers statistical baselines, threshold calibration and drift alerting.
Surrogate Modelling and Physics-Informed AI:
Build and iterate surrogate models in limited-data scientific settings. Implement and validate PINN and physics-constrained approaches designed with the AI Research Scientist, and check physics constraints against engineering requirements.
Active Learning and Experiment Selection:
Implement active learning pipelines and Bayesian experimental design. Configure acquisition functions and integrate them with versioned feature data, so each physical experiment delivers the most value.
MLOps and Deployment:
Package, deploy and maintain models on the team’s ML platform: MLflow, Docker, AWS EKS/Kubernetes, CI/CD and model monitoring. Contribute improvements to the platform under the Sr Technologist’s direction.
Collaboration and Technical Contribution:
Partner with domain scientists and process engineers to turn engineering problems into ML problems. Define feature requirements with the Data Engineer. Take part in design and code reviews, document model decisions to production-handoff standard, and guide junior engineers informally.
Requirements
Education
Bachelor’s, Master’s or PhD in Artificial Intelligence, Machine Learning, Computer Science, Physics, Materials Science, Electrical Engineering, Applied Mathematics or a related field. Equivalent depth shown through open-source contributions or a significant GitHub portfolio will be considered.
Experience
Minimum 3 years of revelent experience
Hands-on ownership of at least two ML models from design through to deployment in product development or production. Experience applying ML to physical, scientific, manufacturing or process development data is a strong differentiator. Formal team leadership not required.
Must-Have Skill Detail
Python - Strong to expert proficiency; primary language for all ML work
PyTorch - Proficient to expert: custom loss functions, independent training and evaluation
Computer Vision - CNN and U-Net, plus ViT or transformer vision, for inspection, defect detection and measurement
Anomaly Detection - Anomaly detection on sensor, time-series or process data
Surrogate Modelling - Data-efficient ML in limited-data scientific settings
Active Learning and Bayesian Methods - Implement acquisition functions, uncertainty-guided experiment selection, or Gaussian process models
Practical MLOps - MLflow, Docker, Git, CI/CD; deploying and monitoring models
Model Evaluation - Ablation studies, confidence calibration, validation methodology
Good to Have
PINNs or physics-constrained models: hands-on implementation
Physics, materials science, semiconductor or process development background
AWS: EKS, S3, SageMaker
Reinforcement learning basics
RAG, LLM fine-tuning (LoRA, QLoRA), LLM gateways (LangFuse, PortKey)