AI/ML Software Engineer, EM and Physics Surrogates
Falcomm
·Yesterday
- Location
- Atlanta, Georgia
- Workplace
- Onsite
- Type
- Full-time
- Department
- Software Engineering
- Experience
- 3+ years
- Education
- Master
- Visa
- Not sponsored
- Source
- Workable
Description
Are you passionate about using machine learning to replace hours of physics simulation with milliseconds of inference? At Falcomm, we are building GaNdalph.ai, an AI-native RF chip design platform that turns design cycles of months into minutes. We are seeking a full-time AI/ML Software Engineer to build and ship the models that predict electromagnetic and thermal behavior of on-chip structures. Your models will stand in for slow field solvers inside the design loop, so the platform can explore thousands of layouts where a human team could try a handful. This role sits at the intersection of software engineering, machine learning, and applied physics.
We are seeking an engineer who thrives in a fast-paced environment, cares about measurable accuracy, and is motivated to ship working product. This is a product role: your models decide the geometry of real chips.
Responsibilities
- Surrogate Models: Build, train, and ship models that predict the electromagnetic response of passives such as inductors, transformers, baluns, and matching networks directly from layout geometry.
- Dataset Generation: Automate large-scale simulation campaigns with EM solvers to create training data across foundry metal stacks.
- Inverse Design: Develop optimization and generative methods that go from a target response to a manufacturable layout.
- Thermal Modeling: Extend the surrogate approach to heat flow in high-power-density designs.
- Accuracy and Trust: Quantify model error and uncertainty, and automatically fall back to full simulation when a prediction is outside its trusted range.
- Production Integration: Package and serve models for fast inference inside the GaNdalph.ai design flow.
- Cross-Functional Collaboration: Partner with EM design engineers and layout engineers to turn design rules and physical constraints into deployable software.