Engineering, Sr Engineer

Job Description and Requirements

We Are

Synopsys is the leader in engineering solutions from silicon to systems, enabling customers to rapidly innovate AI-powered products. We deliver industry-leading silicon design, IP, simulation and analysis solutions, and design services. We partner closely with our customers across a wide range of industries to maximize their R&D capability and productivity, powering innovation today that ignites the ingenuity of tomorrow.

You Are

You have spent years building ML systems that work at scale, not just in notebooks but in production, where data pipelines break, training runs fail at hour 47, and someone has to own the fix. The difference between a model that ships and one that stays in research is usually the unglamorous work of versioning datasets, writing tests, and making sure the thing can actually be called from a product workflow. You are the engineer who closes that gap.

Scientific simulation data is different. Irregular 3D meshes, physics constraints that matter, and datasets where one bad simulation can poison a training run. You have worked with data like this before, or you are ready to learn why it matters. Either way, you know that the model is only as good as the data it was trained on, and you care about that upstream problem as much as the architecture.

Deployment is not someone else's job. You think about what happens after the checkpoint: how it gets packaged, how inference gets served, how a product engineer in a different timezone calls it without breaking their workflow. At Synopsys, you will build the infrastructure that makes Large Physics Foundation Models real for CFD, thermal, and electromagnetics products that customers actually ship.

What You'll Be Doing

- Build and extend dataset factories that produce repeatable, versioned corpora across geometry, mesh, and solver setup using agentic orchestration, shared across CFD, Icepak (electronics cooling), and HFSS (high-frequency electromagnetics)
- Produce large, quality-gated datasets for training Large Physics Foundation Models, ensuring bad simulations are quarantined and data provenance is tracked end to end
- Take trained checkpoints through deployment: package models, stand up inference serving, and build integrations that product teams can call inside their workflows for initialization, prediction, or design-loop acceleration
- Partner with product teams to understand deployment constraints, test in real workflows, and support forward-deployed engineering when models go live
- Write production-grade Python with testing, versioning, CI, and operability for long-running jobs that span data generation, training, and serving
- Build data loaders and training pipelines for irregular 3D simulation data using PyTorch, handling distributed training, checkpointing, and physics-based evaluation metrics
- Own the ambiguous boundary between data engineering, model training, and product deployment, closing gaps instead of waiting for a perfect handoff

The Impact You Will Have

- Enable Synopsys simulation products to shift from "run a solver for every design" to "a model trained on high-fidelity data proposes a physically useful field in seconds," fundamentally changing how customers design cars, aircraft, chips, packages, and RF systems
- Build the data and deployment infrastructure that competitors cannot replicate from a research paper, giving Synopsys a durable advantage in physics-informed ML
- Accelerate design loops for customers by orders of magnitude, letting them explore more configurations, fail faster, and use expensive solvers only where fidelity still matters
- Create reusable dataset factories that scale across multiple physics domains, so campaigns can generate trusted multi-physics corpora without reinventing the pipeline every time
- Make Large Physics Foundation Models operationally real for product teams, so they can ship models that customers trust and that survive contact with production workflows
- Establish reproducible training runs and physics-based evaluation standards that ensure models fail gracefully and teams know the bounds of what they can and cannot predict
- Support product integrations that let downstream engineers call your models without needing to understand the ML stack, just the physics constraints that matter for their workflow

What You'll Need

- Master's degree in Mechanical Engineering, Aerospace Engineering, Electrical Engineering, Applied Mathematics, Physics, or a related field (PhD welcome but not required)
- 2-3+ years building ML systems in production, with meaningful time spent on large scientific datasets, simulation data, HPC environments, or deploying models beyond notebooks
- Strong Python engineering habits: testing, versioning, CI/CD, API design, and operability of long-running jobs
- Hands-on deep learning experience with PyTorch or equivalent, including data loaders for irregular 3D data, distributed training, and checkpointing
- Experience with transformers, vision-language models, neural operators, geometric deep learning, graph networks, or physics surrogates is a strong plus
- Demonstrated ability to own work across ambiguous boundaries between data engineering, model training, and product deployment

Who You Are

- You close gaps instead of waiting for someone else to define the interface. If the data pipeline does not exist, you build it. If the deployment story is unclear, you figure it out with the product team.
- Scientific honesty matters to you. You quarantine bad simulations, report what the model does not predict, and own the bounds of failure instead of overselling what works.
- You defer to specialists on what you do not know and collaborate effectively with product engineers who will live with your deployment long after you hand it off.
- You can explain a technical tradeoff to a product manager in two sentences without losing the nuance, and you can debug a distributed training failure at 3am when the run is 80% done.
- You think in systems, not just models. The checkpoint is not the finish line. Packaging, serving, versioning, and supportability are part of the job, and you treat them that way.

The Team You'll Be Part Of

The team's job is to make Large Physics Foundation Models real for CFD, Icepak, and HFSS. Research defines the architectures and learning objectives. This engineering seat makes the data and the runtime real. You will work with specialists who understand the physics deeply, product teams who know what customers need, and other ML engineers who care about building systems that scale. The core mission has three parts: reuse and scale dataset factories so campaigns produce large, clean, 3D physics corpora; train large models on that data with reproducible runs and physics-based evaluation; and deploy and support those models so product teams can integrate them, operate them, and trust their limits.

Rewards and Benefits

We offer a comprehensive range of health, wellness, and financial benefits to cater to your needs. Our total rewards include both monetary and non-monetary offerings. Your recruiter will provide more details about the salary range and benefits during the hiring process.

---

At Synopsys, we want talented people of every background to feel valued and supported to do their best work. Synopsys considers all applicants for employment without regard to race, color, religion, national origin, gender, sexual orientation, age, military veteran status, or disability.