Software Engineer, Perception Autolabeling

Salary: $176000 - 195000 per year Requirements: We are looking for someone with 3+ years of backend or data pipeline engineering experience. We need strong Python skills, with C++ familiarity preferred. We value experience handling large datasets with PySpark or comparable tools. We expect a solid grasp of machine learning fundamentals, including model inference, embeddings, and evaluation metrics. We are seeking experience integrating foundation models such as Gemini, OpenAI, or Anthropic in production environments at scale. Responsibilities: We will build and run the autolabeling pipeline to speed up vehicle attribute classification work. We will develop observability tools to track system performance, latency, and failure patterns. We will design and execute experiments, then analyze results to improve labeling workflows. We will integrate the pipeline with existing systems and partner closely with data infrastructure teams. We will document architecture, prepare runbooks, and support a smooth project handoff at completion. Technologies: Backend Support Machine Learning Python PySpark More: We are a leader in autonomous vehicle technology, and this role joins our Perception Attributes Autolabeling Pipeline team in Foster City, CA, with onsite work required at least three days per week. This position offers a pay range of $75-$83 per hour plus benefits. We provide a collaborative environment where you can work on cutting-edge perception systems, contribute to real-world transportation innovation, and grow alongside talented ML engineers and system architects. Eligible consultants receive a comprehensive benefits package that includes medical, dental, and vision coverage, health savings and flexible spending accounts, supplemental life insurance, short- and long-term disability, business travel insurance, 401(k) with match, and weekly pay. We are part of ManpowerGroup, a global workforce solutions company with a long-standing reputation for helping organizations and candidates succeed across more than 75 countries. last updated 26 week of 2026