YOUTUBE
Tech Videos
Sources AI Engineer All-In Podcast Andrej Karpathy Anthropic Apple Apple Developer AWS Events ByteByteGo Computerphile Cursor Dwarkesh Patel EO Fireship GitHub Google Cloud …
Sources
- AI Engineer
- All-In Podcast
- Andrej Karpathy
- Anthropic
- Apple
- Apple Developer
- AWS Events
- ByteByteGo
- Computerphile
- Cursor
- Dwarkesh Patel
- EO
- Fireship
- GitHub
- Google Cloud Tech
- Google DeepMind
- Google for Developers
- Hung-yi Lee
- Lenny's Podcast
- Lex Clips
- Lex Fridman
- Life at Google
- Marques Brownlee
- Microsoft
- No Priors: AI, Machine Learning, Tech, & Startups
- Numberphile
- NVIDIA
- OpenAI
- Perplexity
- Quanta Magazine
- Slack
- The Pragmatic Engineer
- Visual Studio Code
Watch First
How to access your Copilot CLI session from anywhere is the single most practical watch today, demonstrating how a simple --remote flag can instantly bridge a live, context-aware terminal session to any device’s browser without requiring clunky VM or remote desktop setups.
Highlights by Theme
Developer Tools & Platforms
GitHub demoed a highly practical --remote flag for their Copilot CLI that securely exposes a live terminal session via URL or QR code, while another short clip shows the CLI summarizing entire undocumented codebases in seconds. Meanwhile, OpenAI highlighted a testimonial claiming Wasmer used Codex to build an edge-ready JavaScript runtime in C++ in just over two weeks, relying on the model to catch subtle bugs in a language their core Rust team was less familiar with. Additionally, AWS events outlined upcoming enterprise orchestration features with an AgentCore Registry and a DevOps Agent aimed at automated root-cause analysis.
AI & Machine Learning
NVIDIA released “Ising,” a substantive family of open AI models purpose-built for quantum computing, which includes a Vision Language Model for automated hardware calibration and a decoding model optimized for surface code correction. On the foundation model front, Google for Developers dropped a rapid-fire spec sheet for Gemma 4, confirming four distinct variants: Effective 2B, Effective 4B, 26B Mixture-of-Experts, and a 31B Dense model. Finally, No Priors featured a pragmatic discussion arguing that while the technical blueprint for superintelligence will be set within a couple of years, actual deployment will take decades due to the high friction of category-specific post-training.
Hardware & Infrastructure
Marques Brownlee offers a fascinating teardown of LG’s unreleased rollable smartphone, showcasing a functional motorized expanding display whose exposed gears and flexible screen clearly highlight why this over-engineered form factor would fail real-world durability tests. In the telecom space, NVIDIA outlined its AI-RAN concept, envisioning 5G base stations running Metropolis VSS to host real-time AI agents for edge tasks like point-cloud generation and traffic simulation. Other infrastructure topics appearing in the feed include AWS sessions covering Serverless Apache Airflow and NVIDIA Isaac Sim, alongside a largely marketing-focused NVIDIA video celebrating 20 years of CUDA.
Everything Else
Lenny’s Podcast tackles product management in the AI era, arguing that as foundation models massively accelerate software builds, PMs must stop over-indexing on year-long sequential roadmaps and start acting like visionary CEOs. For a complete break from tech, Lex Clips features historian Lars Brownworth discussing the psychological terror of the Vikings, while Numberphile hosted a lighthearted, math-free art experiment. Rounding out the list are historical discussions on how censors inherently miss what matters and a breakdown of the analytics engine behind NFL IQ.