System Modeling (Computation)

<h3>About Unconventional</h3> <p>Since 2022, AI has entered the mainstream, reshaping entire industries from education and software development to fundamental consumer behaviors. This revolution has created an unprecedented demand for computation - a demand that is now fundamentally limited by energy, not just in the datacenter, but at a global scale.</p> <p>At Unconventional, our mission is to solve this. We are rethinking computing from the ground up to build a new foundation for AI that is 1000x more efficient. We're doing this by exploiting the rich physics of semiconductors, mapping neural networks directly to the device physics rather than relying on layers of inefficient abstraction.</p> <h3>The Role</h3> <p>As a Member of Technical Staff, System Modeling (Computation), you will be part of a hands-on R&amp;D team building simulation frameworks that enable evaluation and rapid iteration across all layers of unconventional physics-based computing systems for machine learning workloads. “Extreme co-design” is our guiding principle.</p> <p>System Modeling is a multi-disciplinary effort, and the team we’re building reflects that. The role involves development of physics-based system models, GPU-accelerated ML system simulations, and cross-layer system integration. You don’t need to be an expert in all of these, but you have to be very strong in at least one, and solid in the rest.</p> <h3>Responsibilities</h3> <p>You will be responsible for one or more of the following tasks:</p> <ul> <li><strong>Architecting Foundational Solvers:</strong> Building large-scale, GPU-accelerated, high-fidelity numerical differential equation solvers (ODE, SDE, CDE, PDE). You will build tools that enable rapid iteration, multiple architectures, and rich metrics/visualization, leveraging frameworks best suited for scientific ML (e.g., JAX, PyTorch, or custom CUDA/Triton kernels).</li> <li><strong>Bridging Physics and Machine Learning:</strong> Developing physics-based surrogate models of device- and system-level behavior in unconventional compute. You will create clean, composable abstractions that expose algorithm–hardware tradeoffs and enable cross-layer optimization via end-to-end autodiff.</li> <li><strong>Extreme Co-Design &amp; Collaboration:</strong> Working closely with hardware and algorithm teams to understand their simulation needs, supporting everything from high-level algorithm development to the low-level verification of novel, analog hardware.</li> </ul> <h3>Minimum Qualifications</h3> <ul> <li>Education</li> <ul> <li>MS/PhD in a quantitative field (AI/ML, Computer Science, Physics, Electrical Engineering, Applied Math), or BS with substantial, clear evidence of equivalent research/engineering depth.</li> </ul> <li><strong>Scientific Machine Learning (SciML) &amp; Numerical Methods</strong> <em>(replaces Dynamical computation knowledge)</em></li> </ul> <ul> <li>Deep expertise in numerical differential equation solvers (e.g., ODE, SDE, DDE) and their implementation on parallel architectures (e.g., Rosenbrock methods, Euler-Maruyama, adjoint methods, implicit solvers).</li> <li>Experience with high-performance, customized GPU kernel development for numerical methods, including GPU memory optimization and multi-GPU scaling.</li> <li>Experience building effective neural network surrogate models (e.g., Neural ODEs) for complex dynamical systems.</li> </ul> <ul> <li>ML and systems fluency</li> <ul> <li>Solid understanding of modern AI/ML architectures and training/inference workflows.</li> <li>Strong experience implementing and debugging ML models in PyTorch (preferred) or similar, with practical experience profiling, optimizing, and stabilizing non-trivial large-scale ML systems.</li> </ul> </ul> <h3>Preferred Qualifications</h3> <p>We are looking for well-rounded candidates. While the minimum qualifications focus on the core modeling and ML expertise, candidates who possess the following qualities will be the most impactful.</p> <ul> <li>Software Engineering &amp; API Craftsmanship</li> <ul> <li>Exceptional Python engineering skills with a passion for Developer Experience (DX), elegant API design, strong typing, and composability.</li> <li>Experience with compiler-friendly ML paradigms and internals (e.g., JAX vmap/pmap/jit, PyTorch autograd/torch.compile, custom XLA or Triton kernels).</li> <li>A track record of building open-source tools, scientific libraries, or serious simulation/modeling frameworks from scratch.</li> </ul> <li>Dynamic Systems Knowledge</li> <ul> <li>Familiarity with analog dynamic systems, including transient responses, and nonidealities such as nonlinearity, quantization, random noise, and feedback/stability.</li> </ul> <li>Systems Thinking&nbsp;</li> <ul> <li>Demonstrated ability to reason across multiple layers of the stack: algorithm, software, compiler, runtime, and hardware.</li> <li>Able to cleanly connect model architecture choices to system performance implications (memory bandwidth, communication patterns, latency, energy, and numerical stability).</li> <li>Experience applying efficiency techniques natively within modeling frameworks (quantization, sparsity, pruning, distillation, kernel fusion, etc.).</li> </ul> </ul> <h3>Why Join Us?</h3> <ul> <li><strong>The Mission:</strong> Redefine computing for the next 50 years by solving the fundamental energy limitation of AI at a global scale.</li> <li><strong>The Impact:</strong> Shape the company's future as a foundational team member. Enjoy massive ownership and an outsized opportunity to drive change.</li> <li><strong>The Challenge:</strong> Dive into deeply complex, intellectually stimulating, and unsolved problems at the cutting edge of multiple, converging fields.</li> <li><strong>The Perks:</strong> A comprehensive package including best-in-class health benefits, 401k matching, truly unlimited PTO, and complimentary meals when working from our Palo Alto office.<br><br></li> </ul>