Collection

Library

Browse your videos and channels

Video Assets

919

Tracked Channels

30

Videos 919 Channels 30

Andrej Karpathy

How I use LLMs

107 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: LLMs become reliably useful when treated not as omniscient chatbots but as probabilistic models whose performance depends on deliberate model selection, context manage…

Unread

Andrej Karpathy

The spelled-out intro to language modeling: building makemore

84 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: A character-level bigram language model can be built first as normalized transition counts and then as a one-layer neural network trained with negative log likelihood;…

Unread

Andrej Karpathy

Building makemore Part 3: Activations & Gradients, BatchNorm

91 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Reliable neural-network training depends on actively controlling and diagnosing activation, gradient, and parameter-update scales; principled initialization and normal…

Unread

Andrej Karpathy

Building makemore Part 4: Becoming a Backprop Ninja

88 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Autograd should be used in production, but engineers need to understand and be able to manually derive tensor-level backpropagation to debug neural-network failures, v…

Unread

Andrej Karpathy

Stable diffusion dreams of steampunk brains

1 min read

## At-a-Glance - Verdict: Skip - Core thesis: The transcript contains no substantive argument or usable explanation; it is almost entirely repeated filler phrases. - Why it matters: There is no recoverable insight about …

Unread

Andrej Karpathy

Deep Dive into LLMs like ChatGPT

172 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: LLM assistants are built through a three-stage pipeline—pre-training for broad knowledge, supervised fine-tuning for assistant behavior, and reinforcement learning for…

Unread

Andrej Karpathy

The spelled-out intro to neural networks and backpropagation: building micrograd

105 min read

## At-a-Glance - Verdict: Skim - Core thesis: Modern neural-network training reduces to constructing a differentiable computation graph, applying local derivatives backward via the chain rule, and iteratively updating pa…

Unread

Andrej Karpathy

[1hr Talk] Intro to Large Language Models

51 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Large language models are best understood not as chatbots but as probabilistic, partially inscrutable computing kernels that compress knowledge, follow learned interac…

Unread

Andrej Karpathy

Building makemore Part 2: MLP

55 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: This hands-on lecture replaces a brittle bigram count model with a small embedding-based MLP language model, showing how learned representations, minibatch optimizatio…

Unread

Andrej Karpathy

Let's build GPT: from scratch, in code, spelled out.

89 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Karpathy builds a decoder-only GPT from a bigram baseline into a scaled transformer to show that modern GPT pre-training is architecturally straightforward but operati…

Unread

My First Million

I'm 80 and I wasted 25 years of my life. Don't make my mistake. - Howard Marks

29 min read

## At-a-Glance - Verdict: Skim - Core thesis: Marks argues that durable investment success comes from variant perception, calibrated uncertainty, pre-positioning before crises, and acting despite fear—not from certainty—…

Unread

Every

Granola's Chris Pedregal on Building a $1.5B AI Company

49 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Granola sees meeting notes as a wedge into a larger AI-native work layer: capture uniquely rich meeting context, expose it to any agent, and productize the meeting-adj…

Unread

AI Engineer

Don't Ship Skills Without Evals — Philipp Schmid, Google DeepMind

17 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Treat every agent skill as production code: pair it with outcome-based regression evals, test both triggering and non-triggering cases, and use ablation tests to prove…

Unread

AI Engineer

WTF Is the Context Layer? The Missing Infrastructure for Production Agents — Prukalpa Sankar

14 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Production agents need a shared, portable context layer that converts company knowledge, expertise, norms, and operational history into governed, machine-usable contex…

Unread

AI Engineer

Simon Willison in conversation with Cat Wu & Thariq Shihipar, Anthropic

42 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: As frontier models become capable of sustained autonomous work, the engineering bottleneck shifts from implementation to judgment, orchestration, verification, permiss…

Unread

Uncapped with Jack Altman

Sam Blond on the Future of Sales in an AI-Native World | Ep. 54

58 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: AI-native sales platforms will displace incumbents like Salesforce by disrupting labor instead of just IT budgets, with success hinging on creating demand-rich environ…

Unread

AI Engineer

RLM: Recursive Language Models for Large Codebases - Shashi, Superagentic AI

11 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Recursive Language Models (RLM) solve large codebase context problems by externalizing context management into programmable REPL environments where models write code t…

Unread

AI Engineer

In Code They Act, In Proof We Trust — Erik Meijer, Leibniz Labs

13 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: AI agents with tool-calling capability are fundamentally unsafe until we reify their execution plans as programs that can be formally verified before execution, using …

Unread

AI Engineer

Don't Build Agents You Can't Answer For — Addy Osmani

13 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Engineers must shift from task execution to evidence-based accountability; agents automate work but humans must own decisions, verify output, and answer for production…

Unread

20VC with Harry Stebbings

Wix Founder: Will Base44 Win the Vibe-Coding Wars? | The Truth About the Economics of Vibe-Coding

51 min read

## At-a-Glance - Verdict: Watch fully - Core thesis: Vibe coding will expand software creation but will not soon replace trusted, vertically integrated business systems; Wix is therefore defending its core while using Ba…

Unread