Root Access
Machine Learning Engineer
About this role
Root Access, a NYC-based electronics startup, seeks a Machine Learning Engineer to design physics-informed neural networks that solve electromagnetic and thermal equations for PCB design optimization. You'll build data pipelines converting circuit files into tensor representations and implement sim-to-reality calibration using lab measurements to enable real-time physics predictions for designers.
What you'll do
- Design and train physics-informed neural networks (PINNs, FNOs, Neural Operators) to solve Maxwell's and Helmholtz equations within loss functions
- Develop high-performance data pipelines converting PCB files (ODB++, IPC-2581, STEP, Gerber) into tensor grids, SDFs, or graph embeddings
- Implement differentiable physics calibration to ingest VNA, TDR, and EMI measurements for fine-tuning material and manufacturing parameters
- Integrate multi-modal architectures connecting GNNs or LLMs with downstream physics engines
- Optimize training and inference on GPU clusters for sub-100ms forward-pass predictions
- Collaborate with electrical and firmware engineers on frontier electronics applications
What they're looking for
- Scientific Machine Learning (SciML) and physics-informed neural networks
- PyTorch or JAX (4+ years expert level)
- PINNs, DeepONets, Fourier Neural Operators
- Partial differential equations and automatic differentiation
- Spatial data manipulation (NumPy, SciPy, Shapely, Open3D)
- GPU optimization and deep learning frameworks (NVIDIA Modulus, DeepXDE, PyTorch Geometric)
- Numerical optimization (Adam, L-BFGS)
- ECAD file format parsing and geometric processing
Benefits
- Work at a top-tier funded NYC startup in frontier electronics
- Collaborate with experienced engineers across electrical, firmware, and software domains
- Focus on cutting-edge physics-ML research with real-world hardware applications
- GPU cluster resources for training and optimization
Opens the application — the Jobs AI extension fills it for you. Set up autofill
Opens the official application on the employer’s site. No login required.
Root Access
Root Access is a hardware startup developing AI-focused products that combine electronics, firmware, and machine learning. The company is hiring Electronics Engineers, Firmware/Embedded Engineers, and Machine Learning Engineers to build and optimize its hardware and AI systems.
View all jobs at Root AccessLikely interview questions
- Walk us through a PINN or FNO project you've built—how did you encode the physics constraints into the loss function, and what challenges did you face?
- Describe your experience converting geometric or CAD data into tensor representations for neural networks. Which libraries and approaches have you found most effective?