Machine Learning Engineer, Zoox

• Extend triagers from simulation to fleet/service data
• Strong Python / Pyspark / SQL skills
• Coursework in machine learning, computer vision, or NLP. Understanding of supervised learning, model evaluation, and common sources of dataset bias and label noise
• Analyze error cases and translate findings into pipeline improvements
• Hands-on experience building production LLM/VLM applications, including agentic workflows, RAG, tool calling, evaluation, and/or fine-tuning
• Familiarity with software-engineering practices such as Git, unit testing, debugging, and code reviews
• Experience evaluating multimodal or vision-language models
• Currently enrolled in a B.S. or M.S. program in a relevant discipline
• Experience with prompt optimization, model fine-tuning, or human-in-the-loop ML systems
• Available to commit to a minimum three-month assignment
• Able to commit a minimum of 40 hours per week
• Able to work on-site at one of our office locations
• Must adhere with Zoox confidentiality requirements, including refraining from using or sharing proprietary company information outside of Zoox, such as in academic research, theses, publications, or presentations

Desirable

• Zoox's part-time student worker program puts you at the center of one of the most ambitious challenges in transportation
• You'll contribute to real projects, work alongside engineers and researchers pushing the boundaries of autonomous technology, and gain experience that goes well beyond the classroom
• We're looking for students who bring strong academic foundations, curiosity that doesn't stop at coursework, and a drive to be part of something that matters
• Support Zoox's Rules of the Road (RotR) program, where a miner + LLM/VLM tooling method automatically mines and triages potential road-rule violation events from simulation and fleet data
• The student will be helping create or improve the data miners that generate the candidate events, help develop and evaluate the VLM triagers — curating golden datasets, measuring precision/recall against expert triage, and extending automation to new regulations and data sources
• Develop and iterate data miners locating potential RotR Violations following SSO requirements
• Develop and iterate on VLM/LLM workflows and pipelines for classifying RotR events
• Curate golden datasets and evaluate triager precision/recall against human triage
• We want to be transparent: this is not an internship
• The Part-Time Student Worker Program is designed to complement your academic experience by providing meaningful, ongoing work alongside your studies
• Rather than participating in a cohort-based program, you'll join a team directly and contribute to real projects with real impact
• While the program does not include structured intern programming or a pathway to full-time employment, it offers valuable opportunities to learn, develop new skills, and gain hands-on experience in a professional environment