newsfilter
← Interviews
Sequoia Capital10 min

How to Scale AI Application Inference 100x ft. Fireworks’ Lin Qiao

Lin Qiao

Conference Presentation

Fireworks addresses the critical gap between generic model capabilities and specific application needs by co-optimizing inference across quality, speed, and concurrency for over 100,000 variables. This strategic approach reduces inference costs by 10x to 100x through virtual cloud infrastructure that dynamically selects hardware and aligns data distributions, enabling rapid enterprise scaling. The platform has already demonstrated success in transforming high-cost operations into sustainable models, exemplified by a food chain expanding to 1,000 shops and a software firm serving 25 million developers within three months.