Latent Space: The AI Engineer Podcast
swyx + Alessio
The podcast by and for AI Engineers! In 2024, over 2 million readers and listeners came to Latent Space to hear about news, papers and interviews in Software 3.0.We cover Foundation Models changing every domain in Code Generation, Multimodality, AI Agents, GPU Infra and more, directly from the founders, builders, and thinkers involved in pushing the cutting edge. Striving to give you both the definitive take on the Current Thing down to the first introduction to the tech you'll be using in the next 3 months! We break news and exclusive interviews from OpenAI, Anthropic, Gemini, Meta (Soumith Chintala), Sierra (Bret Taylor), tiny (George Hotz), Databricks/MosaicML (Jon Frankle), Modular (Chris Lattner), Answer.ai (Jeremy Howard), et al.Full show notes always on https://latent.space
Episodes
Mentioned books
337 snips
Feb 1, 2025 • 1h 9min
The Agent Reasoning Interface: o1/o3, Claude 3, ChatGPT Canvas, Tasks, and Operator — with Karina Nguyen of OpenAI
Karina Nguyen, Research Manager and Post-training lead at OpenAI, shares her impressive journey from working with Claude at Anthropic to creating innovative tools like ChatGPT Canvas. She dives into enhancing human-computer interaction and the collaborative spirit that drives her team. The conversation highlights the importance of user feedback, the complexity of AI behavior, and innovative approaches to improve writing through AI. Karina also discusses the need for trust in AI interactions and the cultural dynamics between AI organizations.
219 snips
Jan 26, 2025 • 1h 16min
Outlasting Noam Shazeer, crowdsourcing Chat + AI with >1.4m DAU, and becoming the "Western DeepSeek" — with William Beauchamp, Chai Research
William Beauchamp, Founder of Chai Research and ex-hedge fund trader, shares insights on building a successful consumer AI company with just 11 people. He highlights their impressive growth, reaching 1.4 million daily active users and over $22 million in revenue. The conversation explores the innovative Chaiverse platform that drastically cuts A/B testing time and enhances user engagement through unique features. Beauchamp also discusses the challenges of data sharing, compliance, and the importance of user feedback in refining AI models.
182 snips
Jan 19, 2025 • 1h
Everything you need to run Mission Critical Inference (ft. DeepSeek v3 + SGLang)
Join Amir Haghighat, co-founder of Baseten, and Yineng Zhang, lead software engineer at Baseten, as they dive into the groundbreaking DeepSeek v3 model. This model boasts 671 billion parameters and has shaken up LLM inference platforms. They unravel the complexities of deploying massive models, discuss the innovations of SGLang, and delve into the challenges of caching technologies. With insights on optimizing AI workflows and a clear manifesto for crucial applications, this conversation is a must-listen for AI enthusiasts!
645 snips
Jan 12, 2025 • 1h 13min
[Ride Home] Simon Willison: Things we learned about LLMs in 2024
Simon Willison, a leading AI blogger, and Swyx, an AI expert, share insights on the evolving landscape of AI in 2025. They discuss exciting advancements in large language models, focusing on cost efficiency and competition challenges. The duo tackles the skepticism around AI agents, emphasizing their limitations and potential applications in various industries. They also dive into the role of AI influencers and the need for credibility in AI-generated content. Finally, they explore the future of AI interfaces and the rising interest in local LLMs and wearable technologies.
380 snips
Jan 10, 2025 • 56min
Beating Google at Search with Neural PageRank and $5M of H200s — with Will Bryk of Exa.ai
Join Will Bryk, CEO of Exa.ai and former SpaceX engineer, as he dives into redefining search technology. Explore how the shift from link-based to semantic search is revolutionizing user experience. Will discusses the power of neural PageRank, hyper-personalized results, and a groundbreaking $5M AI infrastructure. He contrasts traditional search models with a future focused on understanding context and enhancing scalability. Additionally, he offers insights into startup culture and innovative workplace solutions—like nap pods!
94 snips
Jan 4, 2025 • 55min
AI Engineering for Art — with comfyanonymous, of ComfyUI
Discover the fascinating journey of ComfyUI, an innovative open-source tool for AI image generation that challenges traditional interfaces. Unpack the technical intricacies of diffusion models, prompt weighting, and model customization. Learn about the clever design philosophy behind a node execution engine and how it enhances the user experience. Explore the rise of ComfyUI among competitors, community contributions, and future projects, including the integration of new text features and exciting advancements in AI art.
355 snips
Dec 31, 2024 • 1h 52min
Latent.Space 2024 Year in Review
Celebrate the evolution of AI engineering over the past two years. Discover insights into the shift from research to practical applications and the future of AI agents. Dive into discussions around agent collusion, synthetic vs. real data, and the competitive dynamics between major AI players. Explore the latest trends from AI conferences and the impact of memory technologies on machine learning. Reflect on the past year's achievements and exciting prospects for the coming year in AI.
189 snips
Dec 25, 2024 • 49min
2024 in Agents [LS Live! @ NeurIPS 2024]
Graham Neubig, a Professor at CMU and chief scientist at All Hands AI, dives into the future of coding agents. He discusses the rise of agents by 2025, highlighting the outstanding achievements of OpenHands in software engineering. The conversation covers the integration of human expertise into agent functionality, the significance of effective prompts, and the advancements in AI agents within the Sweebench repository. Neubig also tackles challenges in AI development, emphasizing the role of accessible technology and innovative benchmarks for improvement.
67 snips
Dec 24, 2024 • 29min
2024 in Synthetic Data and Smol Models [LS Live @ NeurIPS]
Lubna Ben-Alau, an AI researcher at Hugging Face, dives into the dynamic world of synthetic data and small language models. She discusses how 2024 saw a remarkable surge in synthetic data applications, with notable contributions like Apple's Rephrasing the Web and Hugging Face's Cosmopedia. Lubna emphasizes the transformative impact of synthetic data on model performance and diversity. The conversation also touches on the evolution of small models, highlighting their efficiency, improved privacy, and specialized applications for on-device use.
29 snips
Dec 24, 2024 • 43min
2024 in Post-Transformers Architectures (State Space Models, RWKV) [LS Live @ NeurIPS]
Dan Fu, an AI researcher soon to join UCSD, and Eugene Cheah, CEO of Featherless AI, delve into the future of post-transformer architectures. They discuss innovations like RWKV and state-space models, highlighting their collaborative and open-source nature. The duo examines the challenges of multilingual training and computational efficiency, while also exploring advancements in non-transformer models like Mamba and Jamba. Tune in for insights on scaling models, 'infinite context,' and how new architectures are reshaping the AI landscape!
