Who We Are
Applied Materials is a global leader in materials engineering solutions used to produce virtually every new chip and advanced display in the world. We design, build and service cutting‑edge equipment that helps our customers manufacture display and semiconductor chips – the brains of devices we use every day. As the foundation of the global electronics industry, Applied enables the exciting technologies that literally connect our world – like AI and IoT. If you want to push the boundaries of materials science and engineering to create next‑generation technology, join us to deliver material innovation that changes the world.
What We Offer
Salary: $220,000.00 - $302,500.00
Location: Santa Clara, CA
You’ll benefit from a supportive work culture that encourages you to learn, develop, and grow your career as you take on challenges and drive innovative solutions for our customers. We empower our team to push the boundaries of what is possible—while learning every day in a supportive leading global company. Visit our Careers website to learn more.
At Applied Materials, we care about the health and wellbeing of our employees. We’re committed to providing programs and support that encourage personal and professional growth and care for you at work, at home, or wherever you may go. Learn more about our benefits.
Job Overview
We are seeking a Director‑level Scientific Machine Learning leader to drive the strategy, development, and deployment of next‑generation ML systems for physics- and chemistry-grounded applications , including physics-informed neural networks (PINNs) , operator learning , graph representation learning for molecules/materials (GNNs, equivariant GNNs, Graph Transformers) , and generative/inverse design .
This leader will own critical initiatives that translate cutting‑edge scientific ML research into production‑grade platforms and decision systems delivering measurable outcomes (e.g., accelerated materials discovery cycles, reduced simulation cost, improved experimental yield, reduced time‑to‑formulation , improved reliability of predictions), while building and mentoring high‑performing teams of ML engineers, research scientists, and computational scientists .
The ideal candidate combines deep technical credibility in modern ML and strong grounding in physics/chemistry/materials science workflows , with proven leadership delivering systems end‑to‑end—from problem framing and data strategy to deployment, evaluation, and scientific validation .
Key Responsibilities
- Define and execute the Scientific ML roadmap across:
- Physics-informed learning (PINNs, PDE‑constrained learning, differentiable physics)
- Operator learning (e.g., neural operators for PDE surrogates)
- Graph learning for molecules, crystals, microstructures, and interaction networks (GNNs, equivariant models, Graph Transformers)
- Generative & inverse design (diffusion models, VAEs, language‑guided design, constrained generation)
- Lead and scale a team of applied ML researchers, computational scientists, and ML platform engineers; establish best practices and a culture of scientific rigor, reproducibility, and engineering excellence.
- Partner with R&D, experimental teams, simulation/HPC groups, product/engineering, and leadership to align AI investments with scientific priorities and operational constraints.
- Own resource planning, hiring, performance management, mentorship, and career development, building a high‑output, high‑quality org.
- Identify high‑value opportunities where scientific ML creates step‑change improvements, such as:
- Accelerating materials discovery (candidate screening, property prediction, inverse design)
- Reducing simulation cost via surrogate modeling and emulation (DFT/MD/CFD/FEA surrogates)
- Improving experimental throughput using active learning, Bayesian optimization, and adaptive DoE
- Improving reliability via uncertainty quantification (UQ), calibration, and robust validation
- Drive research‑to‑production translation, ensuring models are:
- Scientifically valid, reproducible, interpretable where necessary
- Maintainable, monitored, and fit for real‑world decision‑making
- Establish robust evaluation frameworks:
- Guard against data leakage, dataset bias, spurious correlations
- Ensure out‑of‑distribution (OOD) robustness and clear model failure modes
- Define scientific validation criteria (agreement with known laws, conservation constraints, experimental confirmation)
- Lead development of:
- Embedding systems and foundation representations for molecules/materials (self‑supervised learning, contrastive learning)
- Drift monitoring, retraining strategies, continual learning with evolving lab/simulation data
- Lead development of physics‑informed & simulation‑aware ML:
- PINNs / PDE‑constrained training for forward/inverse problems, parameter estimation, and boundary‑value problems
- Neural operators (operator learning) and multi‑resolution surrogates for fast approximations of PDE solvers
- Differentiable programming workflows combining ML with simulators (where applicable)
- Lead development of material discovery & molecular/crystal ML:
- Graph neural networks for molecules, crystals, and periodic systems:
- Message passing, Graph Transformers, E(3)/SE(3)-equivariant GNNs
- Heterogeneous graphs linking composition–structure–processing–property data
- Multi‑modal scientific representation learning:
- Text (papers/ELN notes), structured data (composition/process), images (microstructure, SEM/TEM), spectra (XRD, Raman), and simulation outputs
- Generative & inverse design:
- Diffusion/flow/energy‑based models for structure generation and candidate proposal
- Constraint‑aware generation (stability, synthesizability, property targets)
- LLM‑assisted workflows for hypothesis generation, literature mining, and experiment planning (with guardrails)
- Graph neural networks for molecules, crystals, and periodic systems:
- Optimization & closed‑loop experimentation:
- Bayesian optimization, active learning, and multi‑armed bandits for:
- Efficient candidate selection under budget constraints
- Multi‑objective optimization (e.g., performance vs. cost vs. stability)
- Reinforcement learning where appropriate for sequential decision processes (e.g., experimental scheduling, synthesis planning, control)
- Bayesian optimization, active learning, and multi‑armed bandits for:
- Interpretability & scientific trust:
- Set standards for interpretability and scientific explainability:
- Feature attribution, counterfactuals, physics‑consistency checks
- Uncertainty estimation and calibration as a first‑class requirement
- Set standards for interpretability and scientific explainability:
- Data, MLOps, Reliability & Scientific Governance:
- Define data collection/annotation strategy spanning:
- Simulation outputs (DFT/MD/CFD/FEA), experimental results, instrumentation pipelines
- Standards for metadata, provenance, lineage, units, and versioning
- Ensure production Scientific ML systems meet reliability and governance expectations:
- Monitoring, alerting, rollback, privacy/security, documentation
- Reproducibility standards (experiment tracking, data versioning, model cards)
- Partner with platform teams to improve tooling:
- Feature stores, vector databases (for retrieval + scientific context), model registry
- Scalable training/inference pipelines, HPC integrations, workflow orchestration
- Support for multi‑fidelity datasets and distributed compute
- Define data collection/annotation strategy spanning:
- Success Measures (Examples):
- Discovery acceleration: reduced cycle time from hypothesis → validated candidate (e.g., weeks → days)
- Simulation acceleration: surrogate models reducing compute cost or wall‑clock time (e.g., 10–1000× speedups in target regimes)
- Experimental efficiency: fewer experiments per breakthrough via active learning / Bayesian optimization
- Improved prediction quality: better generalization to new chemistries/process conditions; quantified uncertainty and improved decision confidence
- Platform outcomes:
- Lower time‑to‑deploy new models, improved monitoring coverage, faster iteration cycles
- Strong reproducibility and traceability (data/model versioning, audit trails)
- Org outcomes: strong hiring/retention, technical standards, mentorship, cross‑functional credibility
Required Qualifications
- 6–12+ years building and deploying ML systems with demonstrated production and/or scientific impact (industry or research environments).
- 3+ years leading technical teams/projects (manager, tech lead, lead scientist, or equivalent), with a track record of developing senior talent.
- Deep expertise in modern ML and representation learning:
- Transformers, generative models, self‑/semi‑supervised learning
- Strong intuition for generalization, inductive bias, and model failure modes
- Demonstrated experience in Scientific ML, including one or more of:
- Physics-informed neural networks (PINNs)
- PDE‑constrained learning, differentiable physics, operator learning
- Surrogate modeling for scientific simulation
- Strong experience with graph ML for scientific domains:
- Molecular/materials GNNs, Graph Transformers, equivariant models preferred
- Strong proficiency in Python and modern ML frameworks (PyTorch preferred).
- Strong grounding in math/stats:
- Linear algebra, optimization, probability, scientific experimentation / uncertainty
- Ability to communicate complex technical decisions to non‑technical stakeholders and drive alignment across R&D and engineering.
Preferred Qualifications
- Domain depth in one or more:
- Materials science (batteries, catalysts, polymers, semiconductors, alloys)
- Computational chemistry/physics (DFT, MD), continuum modeling (CFD/FEA), multiphysics simulation
- Experience with:
- Uncertainty quantification (UQ), calibration, Bayesian methods
- Active learning, Bayesian optimization, multi‑objective optimization
- Scientific data systems: ELN/LIMS integration, instrument pipelines, data provenance
- Publications, patents, open‑source contributions in scientific ML/materials AI.
- Experience with large‑scale compute and data:
- HPC/GPU clusters, distributed training, Spark/Ray/Dask, workflow orchestration
- MS/PhD in Physics, Chemistry, Materials Science, CS, EE, Applied Math/Stats (or equivalent practical expertise).
Core Competencies
- Scientific ML leadership: sets technical vision, makes pragmatic tradeoffs, delivers scientifically valid and useful systems.
- Scientific rigor: strong evaluation design, recognizes leakage, bias, drift, and OOD failure modes early.
- Product + R&D mindset: translates scientific goals into roadmaps with measurable outcomes and operational adoption.
- Team builder: scales talent while raising standards for reproducibility, quality, and engineering discipline.
- Cross‑functional influence: aligns experimentalists, simulation experts, platform engineering, and leadership.
Technology Stack (Typical)
- Languages/Tools: Python, Jupyter, SQL, Linux/Unix, Git
- ML/DL: PyTorch, PyTorch Geometric, JAX (optional), Scikit‑learn
- Scientific Computing: NumPy/SciPy, CUDA, HPC schedulers (SLURM), distributed training tooling
- Graph & Materials ML: equivariant GNN libraries (as applicable), graph transformers, molecular featurization toolkits
- GenAI: diffusion/flow models, LLM integration patterns, RAG + vector DB (for scientific context), safety/guardrails
- MLOps & Reproducibility: experiment tracking, data/model versioning, model registry, CI/CD, monitoring/drift detection
- Scientific Data: ELN/LIMS hooks, metadata/provenance standards, pipeline orchestration
Additional Information
- Time Type: Full time
- Employee Type: Assignee / Regular
- Travel: Yes, 10% of the Time
- Relocation Eligible: No
- The salary offered to a selected candidate will be based on multiple factors including location, hire grade, job‑related knowledge, skills, experience, and with consideration of internal equity of our current team members. In addition to a comprehensive benefits package, candidates may be eligible for other forms of compensation such as participation in a bonus and a stock award program, as applicable.
- For all sales roles, the posted salary range is the Target Total Cash (TTC) range for the role, which is the sum of base salary and target bonus amount at 100% goal achievement.
Equal Opportunity Employment Statement
Applied Materials is an Equal Opportunity Employer. Qualified applicants will receive consideration for employment without regard to race, color, national origin, citizenship, ancestry, religion, creed, sex, sexual orientation, gender identity, age, disability, veteran or military status, or any other basis prohibited by law.
In addition, Applied endeavors to make our careers site accessible to all users. If you would like to contact us regarding accessibility of our website or need assistance completing the application process, please contact us via e‑mail at , or by calling our HR Direct Help Line at 877‑612‑7547, option 1, and following the prompts to speak to an HR Advisor. This contact is for accommodation requests only and cannot be used to inquire about the status of applications.
#J-18808-Ljbffr