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.
Thomas Wolf, co-founder of Hugging Face, discusses his journey from physics to AI, the power of open-source models to accelerate innovation, the practical challenges of productionalizing AI demos, and why the biggest opportunities for founders now lie in the application layer on top of powerful foundation models.
Daren Fields, Co-Founder & CEO of Virtual Select, discusses the future of manufacturing, emphasizing the role of AI as a tool for human augmentation, not replacement. He explores how to modernize manufacturing by combining a carbon-based workforce with silicon-based systems to prevent defects, reduce costs, and de-risk execution.
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.
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.
Yuki Watanabe from Databricks introduces Evaluation-Driven Development (EDD) as a critical methodology for building production-ready AI agents. This talk explores the five pillars of EDD and demonstrates how MLflow 3.0's new features—including one-line tracing, automated evaluation, human-in-the-loop feedback, and monitoring—provide a comprehensive toolkit to ensure agent quality and reliability.
A detailed look at a multi-model AI architecture for fraud detection that combines the speed of predictive machine learning for structured data with the contextual understanding of encoder LLMs for unstructured data, enabling faster and more accurate real-time decisions.
Stephanie Kirmer, a staff machine learning engineer at DataGrail, adapts her experience as a former professor to address the challenge of evaluating LLMs in production. She proposes a robust methodology using LLM-based evaluators guided by rigorous, human-calibrated rubrics to bring objectivity and scalability to the subjective task of assessing text generation quality.
This talk addresses the unreliability of current AI agents, arguing that prompting is insufficient. It posits that Reinforcement Learning (RL) is the most promising solution, delving into the mechanisms of RLHF and RLVR. The core challenge identified is 'reward hacking', and the discussion explores future directions to overcome it, such as RLAIF, data augmentation, and the development of interactive, online models that can learn in real-time.
Stack Overflow CEO Prashanth Chandrashekar discusses the platform's evolution in the AI era, focusing on licensing its trusted Q&A corpus to major AI labs, expanding beyond Q&A to include discussions and live chat, and the critical role of its enterprise solution in powering internal AI agents. A key insight from their upcoming developer survey reveals that while AI adoption for coding is rising, developer trust in AI-generated output is declining, reinforcing Stack Overflow's position as a vital source of human-curated, reliable knowledge.
Tom Brown, co-founder of Anthropic, shares his journey from a YC founder to a key figure behind AI's scaling breakthroughs. He discusses the discovery of scaling laws that underpinned GPT-3, the mission-driven founding of Anthropic, the surprising success of Claude for coding, and his perspective on what he calls "humanity's largest infrastructure buildout ever."
Get updates straight to your inbox. No spam, just useful content.