{"meta":{"platform":"HackerTube","version":"1.0.0","description":"Community-curated technical videos extracted from Hacker News discussions (2007-2026).","license":"MIT / Open Data","totalVideos":12,"generator":"HackerTube AI & Pipeline","inspiredBy":"Andrej Karpathy (hackertube.org)","updatedAt":"2026-11-03T05:33:35.810Z"},"videos":[{"id":"karpathy-micrograd","youtubeId":"M3B63EGmIbc","title":"The Spelled-Out Intro to Neural Networks and Backpropagation: Building micrograd","channelTitle":"Andrej Karpathy","duration":"2:25:34","durationSeconds":8734,"publishedYear":2022,"topic":"AI & Machine Learning","hnMentions":42,"hnExcerpt":"This is hands-down the single best tutorial on backprop and computational graphs on the entire internet. He codes up an autograd engine from scratch with zero magic.","hnAuthor":"patio11_admirer","hnYear":2022,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=32514104","aiKeyTakeaways":["Step-by-step construction of a DAG scalar autograd engine in ~100 lines of pure Python.","Demystifies backward pass and gradients through clear calculus chain-rule applications.","Trains a 2-layer multi-layer perceptron (MLP) on a binary classification dataset."]},{"id":"bret-victor-principle","youtubeId":"1bUy-1hGZpI","title":"Inventing on Principle","channelTitle":"Bret Victor","duration":"54:19","durationSeconds":3259,"publishedYear":2012,"topic":"Engineering Culture","hnMentions":58,"hnExcerpt":"One of the most life-altering talks any software developer can watch. Changed the way I think about tools, feedback loops, and guiding life principles forever.","hnAuthor":"danso","hnYear":2012,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=3835694","aiKeyTakeaways":["Creators need immediate, tactile feedback on what they are making without cognitive latency.","The difference between living by a craft versus dedicating your life to an uncompromised principle.","Live demonstrations of reactive interactive programming and dynamic procedural generation."]},{"id":"mother-of-all-demos","youtubeId":"yDYCf4bNwN4","title":"The Mother of All Demos, presented by Douglas Engelbart (1968)","channelTitle":"SRI International","duration":"1:40:02","durationSeconds":6002,"publishedYear":1968,"topic":"Retro & History","hnMentions":39,"hnExcerpt":"He unveiled the computer mouse, hypertext, windows, video conferencing, real-time collaborative text editing, and version control in a single 1968 presentation. Staggering.","hnAuthor":"adamnemecek","hnYear":2018,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=18641193","aiKeyTakeaways":["The debut of the graphical user interface, optical mouse, and chorded keyboard.","Concept of dynamic knowledge networks and augmenting human intellect through computing.","Real-time shared collaborative tele-working over early ARPANET prototypes."]},{"id":"karpathy-build-gpt","youtubeId":"kCc8FmEb1nY","title":"Let's build GPT: from scratch, in code, spelled out","channelTitle":"Andrej Karpathy","duration":"1:56:46","durationSeconds":7006,"publishedYear":2023,"topic":"AI & Machine Learning","hnMentions":51,"hnExcerpt":"Takes you from an n-gram character model to a full multi-head causal self-attention Transformer matching the Attention Is All You Need paper. The definitive guide.","hnAuthor":"jeremyphoward","hnYear":2023,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=34415840","aiKeyTakeaways":["Step-by-step implementation of causal self-attention, query/key/value projections, and residual connections.","Layer normalization, position embeddings, and softmax scaling explained intuitively in PyTorch.","Trains a baby Shakespeare-generating GPT live on GPU within under two hours."]},{"id":"rich-hickey-simple-made-easy","youtubeId":"h_L4Rixya64","title":"Simple Made Easy","channelTitle":"Rich Hickey (Strange Loop)","duration":"1:01:28","durationSeconds":3688,"publishedYear":2011,"topic":"Engineering Culture","hnMentions":64,"hnExcerpt":"The gold standard talk on software architecture. Hickey rigorously dissects the difference between 'easy' (familiar, near at hand) and 'simple' (unbraided, unentangled).","hnAuthor":"swannodette","hnYear":2011,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=3145464","aiKeyTakeaways":["Easy is relative to our familiarity and tooling; simple is an objective measure of unentangled concepts.","Complexity arises from intertwining state, time, constructs, and data representations.","Focus on composable, decoupled values rather than encapsulating tangled mutable state."]},{"id":"casey-muratori-clean-code","youtubeId":"tD5NrevFtbU","title":"'Clean' Code, Horrible Performance","channelTitle":"Casey Muratori","duration":"40:53","durationSeconds":2453,"publishedYear":2023,"topic":"Systems & Architecture","hnMentions":33,"hnExcerpt":"Casey demonstrates with real benchmarks how textbook OOP principles like excessive polymorphism, class hierarchies, and virtual functions cause a 15x slowdown on modern hardware.","hnAuthor":"cmuratori","hnYear":2023,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=34964646","aiKeyTakeaways":["Hardware-conscious data layout versus dogmatic object-oriented abstraction layers.","How virtual function dispatch, pointer indirection, and switch flattening impact instruction caches.","Profile-driven optimization achieving an order of magnitude throughput gain with simple switch-case."]},{"id":"3b1b-essence-of-calculus","youtubeId":"WUvTyaaNkzM","title":"The Essence of Calculus, Chapter 1","channelTitle":"3Blue1Brown","duration":"17:04","durationSeconds":1024,"publishedYear":2017,"topic":"Math & Science","hnMentions":27,"hnExcerpt":"Grant Sanderson accomplishes in 17 minutes what standard high school and college professors fail to do in months: build geometric intuition for what an integral and derivative actually mean.","hnAuthor":"minimaxir","hnYear":2017,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=14227702","aiKeyTakeaways":["Finding the area of a circle by unwrapping concentric rings into a right triangle.","Deriving rate of change through infinitesimals without getting bogged down in algebraic formalism first.","Visual geometry as the foundational anchor for advanced mathematical proofs."]},{"id":"patrick-winston-how-to-speak","youtubeId":"Unzc731iCUY","title":"How to Speak - Patrick Winston (MIT)","channelTitle":"MIT OpenCourseWare","duration":"1:03:32","durationSeconds":3812,"publishedYear":2019,"topic":"Engineering Culture","hnMentions":37,"hnExcerpt":"Winston's annual MIT lecture on communication, lecturing, and persuasion. His rules on pacing, the 'Time & Place' rule, and blackboard usage are timeless.","hnAuthor":"georgecmu","hnYear":2019,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=21637955","aiKeyTakeaways":["Your success in life is determined heavily by your ability to speak, write, and the quality of your ideas.","The Uniform Resource rule: cycle through topics and always provide an 'Empowerment Promise' in the first 5 minutes.","The perils of slide text over-density and why physical chalkboard chalk-speed matches human thought."]},{"id":"carmack-quake-graphics","youtubeId":"443UNeGrFoM","title":"Fast Inverse Square Root: A Quake III Algorithm Deep Dive","channelTitle":"Nemean","duration":"14:28","durationSeconds":868,"publishedYear":2021,"topic":"Systems & Architecture","hnMentions":45,"hnExcerpt":"0x5f3759df: The famous magic constant. A masterclass in IEEE 754 floating-point mechanics, bit-level hacking, and Newton-Raphson approximation.","hnAuthor":"kristopolous","hnYear":2021,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=27814231","aiKeyTakeaways":["Deconstruction of bitcasting float representations directly into 32-bit integers.","The mathematical derivation of the magic constant 0x5f3759df minimizing log2 approximation error.","Applying one single iteration of Newton-Raphson method for lightning-fast normal vector normalization."]},{"id":"karpathy-intro-llm","youtubeId":"zjkBMFhNj_g","title":"[1hr Talk] Intro to Large Language Models","channelTitle":"Andrej Karpathy","duration":"1:00:27","durationSeconds":3627,"publishedYear":2023,"topic":"AI & Machine Learning","hnMentions":54,"hnExcerpt":"The clearest 1-hour overview of what LLMs are, how pretraining works, RLHF, and AI safety without any marketing hype or doomer exaggeration.","hnAuthor":"simonw","hnYear":2023,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=38382025","aiKeyTakeaways":["LLMs as a lossy zip file of the public internet compressed into weights via next-token prediction.","The multi-stage pipeline: Pretraining on trillions of tokens, Fine-Tuning (SFT), and RLHF alignment.","Future paradigm of LLMs functioning like a new Operating System kernel (LLM OS orchestrating RAM, disk, and tools)."]},{"id":"bret-victor-future-programming","youtubeId":"8hly31xKli0","title":"The Future of Programming","channelTitle":"Bret Victor (DBX 2013)","duration":"32:45","durationSeconds":1965,"publishedYear":2013,"topic":"Retro & History","hnMentions":48,"hnExcerpt":"Bret dresses up as a 1973 computer scientist giving a talk on what programming will look like in 40 years. He brilliantly exposes how our modern tools are trapped in local optimums.","hnAuthor":"fogus","hnYear":2013,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=5992928","aiKeyTakeaways":["Critique of modern code editing: we still type character streams into sequential text files just like punch cards in the 60s.","Alternative paradigms: spatial representation, goal-directed constraints, and massive concurrent message passing.","The tragedy of adopting the easiest-to-implement solutions and standardizing them into permanent dogmas."]},{"id":"dan-boneh-crypto-intro","youtubeId":"a1pZ9H3v4uY","title":"Zero Knowledge Proofs and Modern Cryptography Foundations","channelTitle":"Stanford Online","duration":"1:18:12","durationSeconds":4692,"publishedYear":2022,"topic":"Security & Hacking","hnMentions":29,"hnExcerpt":"Dan Boneh's Stanford lectures are gold standard. His explanation of interactive proof systems, zk-SNARKs, and homomorphic encryption is unmatched in mathematical rigor.","hnAuthor":"vbuterin_reader","hnYear":2022,"hnDiscussionUrl":"https://news.ycombinator.com/item?id=30912441","aiKeyTakeaways":["Mathematical foundation of interactive versus non-interactive Zero-Knowledge arguments (ZKPs).","Proving knowledge of a pre-image without revealing confidential witness parameters.","Applications to privacy-preserving blockchains, verifiable computation offloading, and cryptographic audits."]}]}