Description
The role owns the training pipeline for models powering Parallel’s search stack and AI agents, including rankers, classifiers, query models, planning, reasoning, and web-task execution. Responsibilities include building production-quality training data from real product usage, fine-tuning and evaluating models, and shipping them safely to traffic used by millions. The position requires deep knowledge of modern models, transformer fine-tuning, data curation, and label quality, along with a bachelor’s degree or equivalent education, training, and professional experience. The team works fully in person between Palo Alto and San Francisco, and the employer offers visa sponsorship.
