From Physics to AI: Building a 200ms Coding Assistant | ft. Guy Gur-Ari, co-founder of Augment Code скачать в хорошем качестве

From Physics to AI: Building a 200ms Coding Assistant | ft. Guy Gur-Ari, co-founder of Augment Code

💛 Augment Code signup link: https://fff.dev/augment Socials: https://bsky.app/profile/mpj.fff.dev https://bsky.app/profile/augmentcode.com https://x.com/guygr https://x.com/augmentcode Join MPJ for an in-depth conversation with Guy Gur-Ari, co-founder of Augment Code and former Google AI researcher. Guy shares how his fascinating journey from theoretical physics to AI led him to build one of the fastest coding assistants in the industry, achieving 200ms response times while maintaining full codebase awareness. ⏰ Quick jump 00:00 - Cold opener with Guy Gur-Ari quote on hallucination problems 01:07 - What makes Augment different: Full codebase context 02:01 - For engineering leaders evaluating coding assistants 02:32 - Introduction to Guy Gur-Ari (PhD, ex-Google, Augment co-founder) 04:28 - Understanding how Augment handles large codebases 06:11 - The technical challenges of building a fast code assistant 09:19 - How Augment achieves 200-300ms response times 14:33 - Guy Gur-Ari's journey from physics to machine learning 16:35 - Working at the IAS (Institute for Advanced Study) 19:20 - The Google years: From physics to AI research 21:41 - The impact of GPT-3 on AI development 24:40 - Why reasoning capabilities matter in AI 27:41 - The importance of direct user feedback 31:23 - Why Augment chose usage-based pricing over seat-based 33:28 - How have generative models changed your workflow? We dive into the technical challenges of building AI that truly understands complex codebases, explore why current AI coding tools often fall short, and discuss how Augment's unique approach to developer tooling emerged from years of machine learning research at Google. Guy explains how his background in physics influenced his perspective on AI development, including his time at the prestigious Institute for Advanced Study (where Einstein once worked). We also explore the impact of GPT-3 on the field and why reasoning capabilities are crucial for the future of AI coding assistants. Whether you're a developer interested in AI tooling, an engineering leader evaluating coding assistants, or just curious about the intersection of physics and artificial intelligence, this conversation offers unique insights into the future of software development.