The MAD Podcast with Matt Turck Why This Ex-Meta Leader is Rethinking AI Infrastructure | Lin Qiao, CEO, Fireworks AI
11 snips
Mar 27, 2025 
Lin Qiao, CEO of Fireworks AI and former leader at Meta, shares insights on revolutionizing AI infrastructure. She reflects on her transition from managing hundreds of engineers to simplifying AI deployment for developers. Lin delves into the evolution of PyTorch, emphasizing its role in democratizing AI, and discusses the innovative strategies behind optimizing AI performance. Topics include function calling, real-time application integration, and future predictions for AI agents in 2025, highlighting the significance of tailored models and competitive data strategies.
AI Snips
Chapters
Transcript
Episode notes
PyTorch's Journey
- Lin Qiao worked on PyTorch at Meta during its transition to AI-first.
- Initially unifying Meta's AI frameworks seemed impossible, but PyTorch's simplicity ultimately won.
Simplicity Scales
- PyTorch's simplicity made it easier for researchers to adopt, leading to wider industry use.
- This open-source approach fueled innovation by enabling sharing of model architectures.
Generative AI's Impact
- Traditional machine learning required companies to build models from scratch, limiting access.
- Generative AI allows building upon foundation models, shifting the complexity to deployment.