Zillow

Principal Applied Scientist role focused on RL post-training for production models. Leads design and deployment of learning systems to shape model behavior, aligning with user value, safety, and business objectives. Involves supervised fine-tuning, preference modeling, RL-based alignment (RLHF, RLAIF, DPO), reward model development, and translating data into training signals. Partners with model/platform teams on training/evaluation efficiency and mentors other scientists.