Today we are excited to introduce 𝗥𝗼𝗯𝗼𝘁 𝗟𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝗶𝗻 𝗜𝗻𝗱𝘂𝘀𝘁𝗿𝘆, a new Neuracore original series exploring how machine learning is transforming robotics across every major industry. In this series, we speak directly with the engineers, researchers, and technical leaders deploying ML in the physical world. From factory floors to warehouses, from hospitals to energy sites, and across any environment where robots must operate with real-world variability. To launch the series, we sat down with someone building that future from the inside. @NimaGard is the Director of AI at @PathRobotics, the company pioneering autonomous welding systems and developing one of the first foundation models for industrial manufacturing. Welding is one of the hardest problems in robotics. The variability, the precision, the real-world messiness all of it breaks most automation systems. And yet Path believes that with the right data, the right models, and the right hardware, welding can become the proving ground for true physical AI. Nima is leading a team training large-scale models on massive, unsupervised datasets collected from robots deployed across global factory floors. They are building custom scanning hardware. They are unifying data pipelines from worldwide fleets. They are pushing machine learning into places where failure is not theoretical, it is molten metal. So we asked him: • Why welding is the perfect proving ground for industrial foundation models • The importance of high-fidelity scanning before, during, and after welds • How every data point, good or bad, strengthens the model and accelerates improvements • The technical infrastructure required to train at production scale across a global fleet • How their methods extend far beyond welding to other core manufacturing workflows Teams across robotics-driven industries are searching for clarity about what machine learning can truly deliver today. This series exists to provide exactly that clarity. youtu.be/caqlk8LCK7Q?si=QR5X…