All Videos

AI Engineer

Maxime Labonne explains that small language models (350M–24B parameters) for edge deployment face unique architectural and training challenges distinct from simply scaling down large models, requiring specialized solutions like short convolutions, massive over-training, and targeted reinforcement learning to overcome memory constraints and 'doom looping' while excelling at agentic tool use.

freeCodeCamp.org

This documentary covers Stanford's Tree Hacks 2026, an elite hackathon where 1,000 students selected from 15,000 applicants compete for $500,000 in prizes sponsored by major AI companies. Participants showcase advanced multi-agent systems, local-first AI tools, and cross-device platforms while sharing strategies on admission, multi-track prize targeting, and rapid prototyping.

New York Times Podcasts

Wirecutter gift expert Samantha Shea joins hosts Kyra Blackwell and Rosie Garen to share curated Mother's Day gift ideas organized by life stage, emphasizing that the best presents acknowledge maternal labor, provide genuine breaks, and avoid adding to a mother's mental load.

In Good Company (Nicolai Tangen)

Jamie Dimon explains that JPMorgan's 200-year success stems from a culture built through daily discipline, relentless anti-bureaucracy tactics like small empowered teams and rigorous follow-up, and zero tolerance for self-serving employees who prioritize process over client outcomes.

The David Lin Report

Economist Clem Chambers predicts 7-9% inflation for 5-10 years as governments print money to fund massive industrial buildouts and AI infrastructure, warning that bond crises are inevitable but manageable through debt monetization.

My First Million

Renee DiResta traces her decade-long investigation into information warfare from early vaccine misinformation to Russian election interference, explaining how analysts now use the 'Actors, Behaviors, Content' framework to identify coordinated inauthentic networks rather than adjudicating specific speech, while detailing the shift from adversarial to collaborative relationships between researchers and tech platforms.

Stanford Online

This lecture details how to scale language model training across massive clusters using 4D parallelism, contrasting TPU and GPU networking architectures while addressing the critical memory bottlenecks—particularly optimizer states—that dominate training costs at scale.

Stanford Online

This lecture introduces distributed training for large language models, explaining how to scale beyond single-GPU memory limits using collective communication primitives (all-reduce, all-gather, reduce-scatter) across hardware topologies ranging from NVLink-connected single nodes to InfiniBand-linked multi-node clusters.

Stanford Online

This lecture bridges the gap between GPU programming abstractions and hardware realities, explaining how thread hierarchies, memory systems, and hardware constraints like warps and bank conflicts determine kernel performance for deep learning workloads.

Stanford Online

This lecture explains why high-dimensional pixel space (approximately 1 million dimensions for standard images) is computationally intractable for diffusion models, and how Variational Autoencoders (VAEs) solve this by compressing images into structured latent spaces that follow standard normal distributions, enabling efficient and meaningful generation.

Altcoin Daily

A high-ranking US military commander revealed the Department of Defense is actively running Bitcoin nodes, signaling a major policy pivot toward viewing proof-of-work technology as critical cybersecurity infrastructure rather than merely a monetary asset, with specific applications for securing military communications against Chinese cyber threats.

Bloomberg Technology

OpenAI-linked stocks plummeted after reports that the AI giant missed its internal sales and user growth targets, raising questions about whether massive infrastructure investments can be justified by current revenue growth rates.