One topic. Every takeSeek and you shall find

Reading up on Andrej Karpathy

30 deep · digging since nov 24, 25

LLMs are complicated now – Ian’s Blog
Modern LLMs have grown complex with many attention variants and mixture-of-experts, echoing the messy evolution of recommendation systems.
Zen and the Art of Machine Learning Research
Success in machine learning research hinges on temperament—persistence, equanimity, and beginner's mind—more than on raw talent or intelligence, akin to Zen practice.
How the Open Knowledge Format can improve data sharing
Google Cloud introduces the Open Knowledge Format (OKF), an open specification for representing curated knowledge as markdown files with YAML frontmatter to enable portable, agent-readable context.
Karpathy's Autoresearch found a 3-year-old bug in our query engine (and improved performance by 11%) - PostHog
PostHog used a Karpathy-style AI agent to find a 3-year-old ClickHouse bug where `toTimeZone()` disabled primary key usage, and the fix cut query time by 37%
LLM Wiki v2 — extending Karpathy's LLM Wiki pattern with lessons from building agentmemory
Extends Karpathy's LLM Wiki pattern with memory lifecycle, knowledge graphs, hybrid search, and automated quality controls to keep personal wikis healthy as they scale.
Show HN: How LLMs Work – Interactive visual guide based on Karpathy's lecture
An interactive visual guide explains how LLMs are built from internet text to conversational assistants, based on Karpathy's lecture.
Thoughts on slowing the fuck down
Rapid adoption of AI coding agents in production codebases leads to compounding errors, complexity, and loss of maintainability; the author advocates for slowing down and maintaining human oversight.
The End of Coding: Andrej Karpathy on Agents, AutoResearch, and the Loopy Era of AI - YouTube
Karpathy argues the role of human coding will shrink dramatically as autonomous AI agents progress from chaining tool calls to independently designing and running experiments.
Autoresearch on an old research idea | Blog
Using Karpathy's Autoresearch loop with Claude Code, the author reduced mean rank by 54% on an old research project, primarily through bug fixes and hyperparameter tuning.
'The Karpathy Loop': 700 experiments, 2 days, and a glimpse of where AI is heading
Andrej Karpathy's 'autoresearch' AI agent ran 700 experiments in 2 days, discovering 20 optimizations that sped up language model training by 11%.
US Job Market Visualizer
A treemap visualizes 342 US occupations representing 143M jobs, using BLS data and LLM-generated AI exposure scores, but the methodology draws skepticism.
Microgpt | Hacker News
Andrej Karpathy's microgpt is a 200-line pure Python file that implements the full algorithmic core of a GPT, including training and inference, with all other complexity framed as mere efficiency.
Inference is the New Sales & Marketing Spend
AI-native startups like Cursor and Lovable treat high inference costs as a growth investment that replaces traditional sales and marketing spend.
You’re Not Taking On Enough Tech Debt
Because LLMs drastically lower the cost of cleaning up messy code, engineers should take more intentional technical debt now and trust future models to service it.
AddyOsmani.com - Agentic Engineering
The article argues for the term 'agentic engineering' to describe disciplined, human-oversight-driven AI-assisted development, contrasting it with the reckless 'vibe coding' approach.
The 80% Problem in Agentic Coding - by Addy Osmani
Addy Osmani argues that while AI agents now write 80% of code for early adopters, this shifts problems to "comprehension debt," where developers understand less of their own codebases.
Andrej Karpathy (@karpathy): "A few random notes from claude coding quite a bit last few weeks. Coding workflow. Given the latest lift in LLM coding capability, like many others I rapidly went from about 80% manual+autocomplete coding and 20% agents in November to 80% agent coding and 20% edits+touchups in December. i.e. I really am mostly programming in English now, a bit sheepishly telling the LLM what code to write... in words. It hurts the ego a bit but the power to operate over software in large "code actions" is just too net useful, especially once you adapt to it, configure it, learn to use it, and wrap your head around what it can and cannot do. This is easily the biggest change to my basic coding workflow in ~2 decades of programming and it happened over the course of a few weeks. I'd expect something similar to be happening to well into double digit percent of engineers out there, while the awareness of it in the general population feels well into low single digit percent. IDEs/agent swarms/fallability. Both the "no need for IDE anymore" hype and the "agent swarm" hype is imo too much for right now. The models definitely still make mistakes and if you have any code you actually care about I would watch them like a hawk, in a nice large IDE on the side. The mistakes have changed a lot - they are not simple syntax errors anymore, they are subtle conceptual errors that a slightly sloppy, hasty junior dev might do. The most common category is that the models make wrong assumptions on your behalf and just run along with them without checking. They also don't manage their confusion, they don't seek clarifications, they don't surface inconsistencies, they don't present tradeoffs, they don't push back when they should, and they are still a little too sycophantic. Things get better in plan mode, but there is some need for a lightweight inline plan mode. They also really like to overcomplicate code and APIs, they bloat abstractions, they don't clean up dead code after themselves, etc. They will implement an inefficient, bloated, brittle construction over 1000 lines of code and it's up to you to be like "umm couldn't you just do this instead?" and they will be like "of course!" and immediately cut it down to 100 lines. They still sometimes change/remove comments and code they don't like or don't sufficiently understand as side effects, even if it is orthogonal to the task at hand. All of this happens despite a few simple attempts to fix it via instructions in CLAUDE . md. Despite all these issues, it is still a net huge improvement and it's very difficult to imagine going back to manual coding. TLDR everyone has their developing flow, my current is a small few CC sessions on the left in ghostty windows/tabs and an IDE on the right for viewing the code + manual edits. Tenacity. It's so interesting to watch an agent relentlessly work at something. They never get tired, they never get demoralized, they just keep going and trying things where a person would have given up long ago to fight another day. It's a "feel the AGI" moment to watch it struggle with something for a long time just to come out victorious 30 minutes later. You realize that stamina is a core bottleneck to work and that with LLMs in hand it has been dramatically increased. Speedups. It's not clear how to measure the "speedup" of LLM assistance. Certainly I feel net way faster at what I was going to do, but the main effect is that I do a lot more than I was going to do because 1) I can code up all kinds of things that just wouldn't have been worth coding before and 2) I can approach code that I couldn't work on before because of knowledge/skill issue. So certainly it's speedup, but it's possibly a lot more an expansion. Leverage. LLMs are exceptionally good at looping until they meet specific goals and this is where most of the "feel the AGI" magic is to be found. Don't tell it what to do, give it success criteria and watch it go. Get it to write tests first and then pass them. Put it in the loop with a browser MCP. Write the naive algorithm that is very likely correct first, then ask it to optimize it while preserving correctness. Change your approach from imperative to declarative to get the agents looping longer and gain leverage. Fun. I didn't anticipate that with agents programming feels *more* fun because a lot of the fill in the blanks drudgery is removed and what remains is the creative part. I also feel less blocked/stuck (which is not fun) and I experience a lot more courage because there's almost always a way to work hand in hand with it to make some positive progress. I have seen the opposite sentiment from other people too; LLM coding will split up engineers based on those who primarily liked coding and those who primarily liked building. Atrophy. I've already noticed that I am slowly starting to atrophy my ability to write code manually. Generation (writing code) and discrimination (reading code) are different capabilities in the brain. Largely due to all the little mostly syntactic details involved in programming, you can review code just fine even if you struggle to write it. Slopacolypse. I am bracing for 2026 as the year of the slopacolypse across all of github, substack, arxiv, X/instagram, and generally all digital media. We're also going to see a lot more AI hype productivity theater (is that even possible?), on the side of actual, real improvements. Questions. A few of the questions on my mind: - What happens to the "10X engineer" - the ratio of productivity between the mean and the max engineer? It's quite possible that this grows *a lot*. - Armed with LLMs, do generalists increasingly outperform specialists? LLMs are a lot better at fill in the blanks (the micro) than grand strategy (the macro). - What does LLM coding feel like in the future? Is it like playing StarCraft? Playing Factorio? Playing music? - How much of society is bottlenecked by digital knowledge work? TLDR Where does this leave us? LLM agent capabilities (Claude & Codex especially) have crossed some kind of threshold of coherence around December 2025 and caused a phase shift in software engineering and closely related. The intelligence part suddenly feels quite a bit ahead of all the rest of it - integrations (tools, knowledge), the necessity for new organizational workflows, processes, diffusion more generally. 2026 is going to be a high energy year as the industry metabolizes the new capability." | XCancel
LLM agent capabilities crossed a coherence threshold in December 2025, causing a phase shift in software engineering with most coding now done via English instructions.
Claude Code and What Comes Next - by Ethan Mollick
Claude Code, using agentic harness and compaction, autonomously built a working website and game, showing a major leap in AI capability for sustained work.
Auto-grading decade-old Hacker News discussions with hindsight
An LLM retroactively grades decade-old Hacker News predictions, sparking debate on hindsight bias, surveillance, and the value of public web archives.
It's Hard to Feel the AGI
Leading AI researchers argue that current LLM scaling will hit a ceiling, requiring new insights before AGI emerges, pushing timelines back 5-20 years.
Thread by @karpathy on Thread Reader App – Thread Reader App
Andrej Karpathy argues that the space of possible intelligences is vast, with animal intelligence being just one specific point, challenging common intuitions.

Takes