An expert overview of post-training techniques for language models, covering the entire workflow from data generation and curation to advanced algorithms like Supervised Fine-Tuning (SFT), Direct Preference Optimization (DPO), and Reinforcement Learning (RL), along with practical advice on evaluation and iteration.
This talk introduces Anaximander, a system designed to bridge the gap between traditional, GUI-driven Geographic Information System (GIS) workflows and modern, code-heavy machine learning practices. Anaximander integrates geospatial foundation models directly into QGIS, allowing experts to interactively orchestrate, run, and evaluate models for tasks like semantic segmentation and object detection on satellite imagery.
At Applied Compute, efficient Reinforcement Learning is critical for delivering business value. This talk explores the transition from inefficient synchronous RL to a high-throughput asynchronous 'Pipeline RL' system. The core challenge is managing 'staleness'—a side effect of in-flight weight updates that can destabilize training. The speakers detail their first-principles systems model, based on the Roofline model, used to simulate and find the optimal allocation of GPU resources between sampling and training, balancing throughput with algorithmic stability and achieving significant speedups.
The panel discusses Anthropic's Claude Cowork and the challenge of user trust in AI agents for everyday tasks. They then analyze the Apple-Google partnership to integrate Gemini into Siri, debating its implications for edge AI, privacy, and hardware limitations. Finally, they explore Linus Torvalds' use of AI for "vibe coding," considering its impact on hobbyist programming and entrepreneurship versus the current limitations in producing production-ready software.
Jure Leskovec introduces Relational Foundation Models (RFMs), a new class of models based on graph neural networks that learn directly from raw, multi-table enterprise data. This approach bypasses manual feature engineering, leading to more accurate, faster-to-deploy, and easier-to-maintain predictive models for tasks like churn prediction, fraud detection, and recommendation systems.
a16z co-founders Marc Andreessen and Ben Horowitz discuss the shift to a decentralized media ecosystem, their investment thesis on supply-driven markets, and the transformative impact of AI. They detail the a16z model of leveraging reputation as a core asset to turn inventors into CEOs and explain why AI represents a fundamental reinvention of computing that will unlock unprecedented growth.
Nikesh Arora, CEO of Palo Alto Networks, shares his unconventional journey and leadership philosophy. He provides a masterclass in building a multi-platform company through strategic M&A, explains why founders should sometimes ignore customers, and reveals how to lead with conviction while managing imposter syndrome.
Stewart Butterfield, co-founder of Slack and Flickr, shares the product frameworks and leadership principles that guided his success. He delves into concepts like "utility curves" for feature investment, the "owner's delusion" in product design, and why focusing on "comprehension" is often more important than reducing friction. He also introduces powerful mental models for organizational effectiveness, such as combating "hyper-realistic work-like activities" and applying Parkinson's Law to team growth.
Intuit CEO Sasan Goodarzi discusses the operational playbook for reinventing a 40-year-old company, from its slow transition to SaaS to its early adoption of AI. He shares insights on winning the SMB market by treating small businesses like consumers, building effective channel partnerships, and developing a platform strategy. Goodarzi also details his leadership philosophy, emphasizing that grit and curiosity are more critical than raw talent.
The panel discusses KPMG's 100-page prompt for its TaxBot, debating the future of prompt engineering versus fine-tuning. They also analyze OpenAI's potential move into selling cloud infrastructure, the impressive capabilities of Google's new image model, Nano-Banana, and new AI-powered fan experiences at the US Open.
Stefan Saroiu from Microsoft Research details Project STEMA's six-year journey tackling the DRAM security flaw, Rowhammer. He discusses how academic research kept the industry honest about DDR4 vulnerabilities, the development of their in-DRAM defense, Panopticon, and its evolution into the industry standard PRAC for DDR5, while highlighting that significant challenges and research opportunities remain.
An overview of the new GPT-realtime speech-to-speech model and the general availability of the Real-Time API, detailing its architecture, advanced capabilities like image input and multilingualism, training methodology, and new enterprise-ready features.
Dr. Michael Timothy Bennett challenges conventional AI paradigms, arguing for a new approach inspired by the principles of living systems. He critiques the separation of software and hardware ("computational dualism"), redefines intelligence as efficient adaptation, and offers a novel theory of consciousness as a "tapestry of valence" essential for genuine intelligence.
Ben Lorica and David Talby of 'The Data Exchange' podcast analyze the 2025 AI Governance Survey, revealing a significant gap between AI adoption and mature risk management. While 30% of organizations have models in production, many lack robust governance frameworks, incident response plans, and comprehensive monitoring, often prioritizing speed-to-market over safety and compliance.
Asha Sharma, CVP of Product for Microsoft's AI Platform, shares insights from working with over 15,000 companies building AI. She discusses the shift from "product as artifact" to "product as organism," the rise of post-training as the new competitive moat, and how agents are transforming organizational structures from hierarchies ("org charts") into task-based networks ("work charts").
Get updates straight to your inbox. No spam, just useful content.