<p>Tesla is building robust embodied intelligence through autonomous driving and humanoid robots. Core to reaching this goal is developing intelligent agents that see, move, and speak in the real world. In this role, you'll architect and deploy voice-driven models that allow our agents to listen, understand, and respond in real time powered by our custom AI chip on our embodied products. Your work will be central to making Optimus or our cars as seamless as talking to a friend. Most importantly, you will see your work repeatedly shipped to and utilized by thousands of Humanoid Robots in real world applications and millions of cars.</p> <ul> <li> <p>Build robust, real-time ASR (Automatic Speech Recognition) systems for noisy, dynamic environments</p> </li><li> <p>Train end-to-end speech model fine-tuned for embodied agents and deployable on our custom AI chip</p> </li><li> <p>Develop intent recognition pipelines bridging audio input with semantic understanding and control</p> </li><li> <p>Collaborate closely with our perception and control teams to integrate speech into the full sensorimotor stack</p> </li><li> <p>Deploy models on resource-constrained hardware using optimization techniques</p> </li><li> <p>Strong software engineering practices and is very comfortable with Python and Numpy programming, debugging/profiling, and version control</p> </li><li> <p>Deep understanding of modern ASR architectures and toolkits</p> </li><li> <p>Experience training and fine-tuning large-scale speech models using Pytorch</p> </li><li> <p>Familiarity with real-time audio processing and latency-constrained systems</p> </li></ul>