Gradient Dissent: Conversations on AI

Nikolaus West, CEO and co-founder of Rerun, specializes in revolutionizing how we log and visualize multimodal sensor data for robotics. He discusses the challenges of taking AI from the lab to the real world and how Rerun's innovative data stack solves these issues. Key topics include breakthrough techniques in manipulation, the intriguing blend of reinforcement and imitation learning, and the significance of open-source tools in advancing robotics. Nikolaus also shares insights on the future of consumer robotics and the importance of robust data pipelines.

Cristóbal Valenzuela, Founder and CEO of Runway ML, discusses the groundbreaking Gen 4.5 video AI, which recently topped the Video Arena Leaderboard. He reveals how his small team has outpaced tech giants like Google and Meta through focused research and innovative design. Cristóbal explains the crucial role of 'taste' in model creation, advancements in motion and camera control, and how video models can act as simulation engines for diverse applications beyond just media production.

Tuhin Srivastava, CEO and founder of Baseten, shares insights from his journey in the AI inference space. He discusses the shift from small model serving to focusing on large model production needs, influenced by market shifts like ChatGPT. Tuhin explains the importance of dedicated deployments for custom models, along with optimizing runtime performance. He also highlights challenges in adopting new inference chips and the need for open models over closed APIs. The conversation emphasizes inference's critical role in embedding AI into applications at scale.

Edwin Chen, CEO of Surge AI and a veteran ML engineer from Google, Twitter, and Facebook, dives into the challenges of quality data labeling in AI. He discusses Surge's innovative approach that focuses on high-skill data collection over traditional methods. Chen critiques the notion of inter-annotator agreement, especially in complex tasks like poetry and math. The conversation also touches on the role of high-quality data in advancing artificial general intelligence and the necessary evolution of data practices for future AI breakthroughs.

Arvind Jain, CEO of Glean and former founder of Rubrik, dives into enterprise AI's evolution. He discusses Glean's transformation from a basic search tool to a sophisticated AI platform that enhances workplace efficiency. Arvind reveals insights on early implementation of transformer models and the challenges of ensuring AI accuracy amidst potential hallucinations. He reflects on the significant role of AI agents in personalizing employee support and streamlining internal documentation, envisioning a future where every team has their own AI assistant.

Jarek Kutylowski, the CEO and founder of DeepL, reveals the company's journey from pioneering neural machine translation to constructing custom data centers. He discusses the challenges of AI translation, emphasizing the crucial role of human context for high-quality results. Kutylowski also highlights how small teams can successfully compete with giants like Google. The conversation dives into the evolution of speech translation, complexities in model training, and the importance of curated datasets, offering a fascinating glimpse into the future of global communication.

Thomas Dohmke, CEO of GitHub, shares insights into the future of software development influenced by AI. He details the journey of GitHub Copilot's creation and its role in reshaping developer workflows, emphasizing the balance between human skill and AI capabilities. Dohmke also reflects on leading GitHub through its Microsoft acquisition, which sparked unexpected innovation, and the importance of fostering developer happiness for productivity. He highlights the evolving skill set of developers and the significance of integrating computer science education for future generations.

Martin Shkreli, the notorious former pharma CEO turned software entrepreneur, discusses his dramatic transformation from controversial figure to tech innovator. He candidly shares his reflections on drug pricing controversies and the lessons learned from his prison experience. Shkreli critiques mainstream AI approaches in drug discovery and introduces his finance venture, Godel Terminal. He also delves into the cyclical nature of technology bubbles, comparing the current AI hype to the dot-com era, emphasizing the need for cautious investment strategies.

In this episode of Gradient Dissent, host Lukas Biewald talks with Sualeh Asif, the CPO and co-founder of Cursor, one of the fastest-growing and most loved AI-powered coding platforms. Sualeh shares the story behind Cursor’s creation, the technical and design decisions that set it apart, and how AI models are changing the way we build software. They dive deep into infrastructure challenges, the importance of speed and user experience, and how emerging trends in agents and reasoning models are reshaping the developer workflow.Sualeh also discusses scaling AI inference to support hundreds of millions of requests per day, building trust through product quality, and his vision for how programming will evolve in the next few years.⏳Timestamps:00:00 How Cursor got started and why it took off04:50 Switching from Vim to VS Code and the rise of CoPilot08:10 Why Cursor won among competitors: product philosophy and execution10:30 How user data and feedback loops drive Cursor’s improvements12:20 Iterating on AI agents: what made Cursor hold back and wait13:30 Competitive coding background: advantage or challenge?16:30 Making coding fun again: latency, flow, and model choices19:10 Building Cursor’s infrastructure: from GPUs to indexing billions of files26:00 How Cursor prioritizes compute allocation for indexing30:00 Running massive ML infrastructure: surprises and scaling lessons34:50 Why Cursor chose DeepSeek models early36:00 Where AI agents are heading next40:07 Debugging and evaluating complex AI agents42:00 How coding workflows will change over the next 2–3 years46:20 Dream future projects: AI for reading codebases and papers🎙 Get our podcasts on these platforms:Apple Podcasts: https://wandb.me/apple-podcastsSpotify: https://wandb.me/spotifyYouTube: https://wandb.me/youtubeFollow Weights & Biases:https://x.com/weights_biaseshttps://www.linkedin.com/company/wandb

In a riveting conversation, Christopher Ahlberg, CEO of Recorded Future and former founder of Spotfire, reveals his journey in cybersecurity. He discusses the dark web's evolution, sharing gripping tales of cyber espionage, including the breach of the Electoral Assistance Commission. Ahlberg also highlights how AI shapes defenses and attacks, explores the complex interplay between technology and governance, and dives into the security differences between Telegram and Signal. His insights offer a compelling look into the world of AI-driven intelligence and the future of global security.