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.
In a discussion with Astro Teller, Emily Ma from X (formerly Google X) delves into the multifaceted problem of food waste and the moonshot projects developed to tackle it. They explore Project Delta, an initiative that created an "air traffic control" system for surplus food, and Project Chorus, a broader supply chain moonshot designed to give every object a voice through advanced sensors and software.
Ben Horowitz of Andreessen Horowitz discusses how AI is reshaping venture capital, detailing the firm's strategy for managing partners, verticalizing teams, and evaluating investments in real-time. He breaks down the current AI cycle, arguing that application design and model orchestration are key, and explains why the current market's growth is driven by real demand, not just hype.
Ashwin Rajeeva, cofounder and CTO of Acceldata, details the architecture and philosophy behind their Agentic Data Management (ADM) platform. He explains how the Xlake reasoning engine provides crucial context for AI agents to operate across petabyte-scale enterprise data, enabling capabilities like self-healing data pipelines and automated data quality assurance, while also discussing strategies for leading technical teams in the age of AI.
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.
a16z's Anish Acharya and Kevin Rose discuss how AI is fueling a renaissance in consumer technology. They explore the rise of emotional interfaces and AI companions, the importance of backing 'weird and working' founders, and how the modern AI development stack empowers solo creators to build, prototype, and launch products with unprecedented speed and craftsmanship.
Experts discuss the security implications of Windows 10's end-of-life, the transformative but cautious integration of AI into Security Operations Centers (SOCs), the promise and peril of AI-driven automated code repair, and the ongoing battle against social engineering attacks like payroll fraud.
Introducing ChatGPT Atlas, a new AI-powered browser that integrates ChatGPT at its core to rethink web interaction through conversational interfaces, contextual awareness, and autonomous agents that can perform tasks on the user's behalf.
Blaise Agüera y Arcas presents a computational theory of life and intelligence, arguing they are fundamentally the same. He posits that evolution's complexity arises not from random mutation but from merging and symbiosis (symbiogenesis), a process mirrored in technology and collective intelligence. This functionalist perspective suggests consciousness is a mechanism for cooperation and that AI is not an alien "other" but an extension of humanity's existing collective intelligence.
YC partners discuss go-to-market strategies for AI in legacy industries, balancing growth with defensibility, knowing when to pivot even with traction, tackling hard technical problems, and making key hiring and open-source decisions.
Augusto Marietti, CEO of Kong, shares his remarkable founder story, from arriving in the US with $600 to building a category-leading API infrastructure company. He details the seven-year struggle of the initial company, Mashape, the critical pivot to open-sourcing their internal API engine which became Kong, and the explosive growth that followed. Marietti also provides his vision for the future, where APIs serve as the essential connectivity layer for a new generation of AI agents, transforming how machines interact with the internet.
Get updates straight to your inbox. No spam, just useful content.