Embodied AI Engineer

Title: Embodied AI Engineer About Us: UnitX builds the world's leading physical AI systems to automate repetitive visual tasks in factories. UnitX is a fast-moving startup with a team from Stanford, MIT, Google, and beyond. Since inception, UnitX has deployed 1,000+ mission-critical systems across 190+ of the world's leading manufacturers' production lines. Every year, $15B worth of products go through UnitX's AI inspection system to ensure quality. The Role We are seeking a visionary and driven Embodied AI Engineer to serve as a cornerstone of our intelligent robotics team. In this role, you will own the critical intersection of virtual environments and physical intelligence. You will not only architect the high-fidelity physics simulations that serve as the virtual proving grounds for our fleet, but also design, train, and deploy the advanced machine learning paradigms that learn from them. By controlling this entire sim-to-real lifecycle—seamlessly turning massive amounts of synthetic and real-world data into robust, scalable algorithms—you will empower our robots to generalize their intelligence across a multitude of complex, high-precision manufacturing tasks. Key Responsibilities - Architect Intelligent Control Systems: Design, build, and deploy sophisticated, data-driven control architectures. You will create frameworks that synthesize complex, multi-modal sensor inputs, enabling our robotic systems to make nuanced, real-time decisions in dynamic environments. - Bridge the Sim-to-Real Divide: Champion our core data and simulation strategies. Construct highly accurate, physics-based simulation environments to identify and mitigate real-world failure modes. - Build Robust Data Flywheels: Harmonize massive datasets by seamlessly blending unstructured, messy real-world sensor data with task-oriented synthetic data to fuel the continuous training and refinement of our foundation models. - Pioneer Adaptive Behaviors: Drive the application of advanced machine learning methodologies (such as Reinforcement Learning, Imitation Learning, and Behavior Cloning) to physical systems, developing generalized robotic skills that seamlessly transition across diverse manufacturing applications. - Scale for Global Impact & Edge Deployment: Ensure that all algorithmic solutions are rigorously evaluated, diagnosed, and optimized for rapid deployment and massive scalability. You will deploy these models directly to edge compute devices on our global factory network, maintaining uncompromised precision, low latency, and reliability. Basic Qualifications - Education & Experience: MS or Ph.D. in Computer Science, Robotics, Machine Learning, Artificial Intelligence, or a closely related quantitative field. - Programming Excellence: Strong, production-level coding skills with deep expertise in Python and C++. - Machine Learning Mastery: Deep expertise in developing modern machine learning architectures using leading deep learning frameworks (e.g., PyTorch, JAX, TensorFlow), managing the full lifecycle from data ingestion to deployment. - Simulation Expertise: Proven ability to leverage high-fidelity physics engines (e.g., NVIDIA Isaac Sim, MuJoCo, Drake) to train, evaluate, and validate AI models prior to real-world deployment. - Visionary Problem Solving: A demonstrated track record of taking complex, open-ended physical challenges and translating them into elegant, mathematically grounded algorithmic solutions. Preferred Qualifications - Edge Compute Experience: Familiarity with deploying, profiling, and optimizing ML models for real-time inference on robotic hardware (e.g., NVIDIA Jetson, TensorRT, CUDA). - Advanced Embodied AI: Experience exploring or applying large-scale generative models, Vision-Language-Action (VLA) models, or Foundation Models to robotic planning and control. - 3D/Robotics Toolchains: Proficiency with industry-standard 3D asset and robotics description formats (e.g., USD, URDF, MJCF). Benefits: - Competitive salary & equity