Internship, Software Engineer, AI Inference Co Design (Winter 2027)

Consider before submitting an application: This position is expected to start January 2027 and continue through Spring term (ending approximately April 2026). We ask for a minimum of 12 weeks, full-time (40 hours/week) and on-site, for most internships. Our internship program is for students who are actively enrolled in an academic program. Recent graduates seeking employment after graduation and not returning to school should apply for full-time positions, not internships. International Students: If your work authorization is through CPT, please consult your school on your ability to work 40 hours per week before applying. You must be able to work 40 hours per week on-site. Many students will be limited to part-time during the academic year. Our team puts large neural networks into production for efficient real-time inference on compute-constrained edge devices—CPU, GPU, and Tesla’s custom AI ASIC. You will work at the intersection of machine learning and systems, building the frameworks and infrastructure that train, deploy, and run the networks powering Tesla vehicles and Optimus. The goal is dual: experiment with novel architectures under tight constraints, then ship the production stack that extracts maximum performance-per-watt from Tesla’s edge chips (AI5 and beyond) so Optimus and the vehicle fleet can run state-of-the-art intelligence in real time. This is not a typical internship. You will own production-critical pieces of the stack that decide how fast, how efficient, and how capable the AI inside every future Tesla and Optimus becomes—working at the exact point where cutting-edge neural nets meet the most demanding real-time, power-constrained silicon on the planet, and helping turn the vision of ubiquitous, high-performance edge intelligence into reality.

Responsibilities

Iterate quantization, compression and distillation techniques to maximize edge inference efficiency Build robust, high-performance AI frameworks that lower neural networks onto edge devices with extreme efficiency Build and harden the AI infrastructure used to t fine-tune networks for Tesla AI (vehicles + Optimus) Co-design state-of-the-art neural networks onto Tesla’s custom AI ASIC, relentlessly driving latency to the absolute minimum Push the co-design loop between software and hardware so every circuit on the next-generation edge chips (AI5, AI6, and beyond) is used to its fullest

Requirements

Currently pursuing a degree in Computer Science, Computer Engineering, or a relevant field of study with an expected graduation of 2028 or earlier Proficiency with PyTorch (or another major machine-learning framework) Hands-on experience training and deploying neural networks for real-world AI applications Solid understanding of computer systems and computer architecture Experience with CUDA Python proficiency required; C/C++ is a strong plus but not mandatory