About the role
<p><span style="font-size: 18pt;"><strong>Build the Path Forward</strong></span></p> <p>At Path Robotics, we’re building the future of embodied intelligence. Our AI-driven systems enable robots to adapt, learn, and perform in the real world closing the skilled labor gap and transforming industries. We go beyond traditional methods, combining perception, reasoning, and control to deliver field-ready AI that is risk-aware, reliable, and continuously improving through real-world use.</p> <p>Big, hard problems are our everyday work, and our team of intelligent, humble, and driven people make the impossible possible together.</p> <p>Manufacturing demands exceptionally high performance, reliability, and adaptability. Processes like welding involve fast, complex, and poorly modeled physics that traditional simulators struggle to capture - especially in the long tail of real-world conditions.</p> <p>We are building intelligent robotic systems that learn directly from data by combining neural world models with reinforcement learning. Our goal is to give robots the ability to <strong>learn, predict, and plan</strong> in complex manufacturing environments by replacing or augmenting classical physics simulators with fast, high-fidelity learned ones.</p> <p>We are seeking a Senior Machine Learning Engineer to lead the development of a neural welding simulator - a learned world model that captures the visual and physical dynamics of welding and enables large-scale RL training. This role sits at the intersection of generative modeling, robotics, and applied physics. It is research-heavy by design, while still grounded in production reality.</p> <p>This role can be <strong>located in our Columbus, Ohio Headquarters or Remote.</strong></p> <h2><span style="font-size: 14pt;">What You’ll Do</span></h2> <ul> <li>Build a learned world model of the welding process that predicts future system behavior under robot actions.</li> <li>Develop multimodal neural simulators incorporating signals such as 3D scans, video, thermal data, and electrical measurements.</li> <li>Design, train, and evaluate large-scale generative or dynamics models (e.g., video prediction, latent world models, 3D or spatiotemporal representations) capable of long-horizon rollouts.</li> <li>Collaborate with reinforcement learning engineers by integrating the neural simulator into RL pipelines for policy training and evaluation.</li> <li>Run research tracks in parallel with production development, including hypothesis-driven experimentation and ablation.</li> <li>Partner closely with data and MLOps teams to support scalable training, evaluation, and deployment - while remaining comfortable owning pieces of the stack when needed.</li> <li>Translate