Building Voice AI at Scale: Fine-Tuning, Reinforcement Learning, and Personalization

What does it take to build voice AI that hundreds of thousands of people can use across their everyday applications? Accurate speech recognition is only the beginning, with the real challenge making the output fast, reliable, and personalized. In this session, Lawrence, Co-Founder and CTO of Willow Voice, shares the engineering and product lessons behind building voice AI at scale spanning supervised fine-tuning, reinforcement learning, and personalization.‍The session then examines personalization in-depth: establishing good defaults, learning from user corrections, and turning subjective writing preferences into behaviors a model can follow. Attendees will leave with lessons applicable to building specialized AI products well beyond voice.