Andrej Karpathy
PulseAugur coverage of Andrej Karpathy — every cluster mentioning Andrej Karpathy across labs, papers, and developer communities, ranked by signal.
- 2026-05-19 hiring Andrej Karpathy joins Anthropic to help advance language model capabilities.
15 天有情绪数据
-
Sequoia Capital: AI is a computational revolution, agents to dominate 2026
Sequoia Capital hosted its fourth annual AI Ascent event in San Francisco, gathering over 150 AI founders and researchers. Key themes included AI as a fundamental computational revolution, with predictions that 2026 wil…
-
Zenii compiles documents into local AI wikis for faster, consistent knowledge retrieval
Zenii has released a new local-first AI assistant platform designed to improve how users interact with their documents. Unlike traditional RAG workflows that re-synthesize answers on every query, Zenii compiles knowledg…
-
Tesla seeks high-paid data labelers for FSD and Optimus, emphasizing quality and domain knowledge.
Tesla is aggressively hiring data labelers for its FSD and Optimus projects, offering high salaries up to $138,000 annually for manager roles and providing comprehensive benefits. The company emphasizes the need for hum…
-
AI coding beginners err by skipping specs and trusting code blindly
Beginners often make five key mistakes when using AI for coding, primarily stemming from a lack of clear specifications rather than poor prompting. Studies indicate that AI-generated code is more prone to errors and vul…
-
New 'KTTB' unit proposed to measure contrarian AI thinking in seconds
A new unit of measurement, the KTTB (Karpathy-Tweet-To-Backlog), has been proposed to quantify contrarian independent thinking within the AI field. This unit is measured in seconds and is named in reference to Andrej Ka…
-
AI Retrospective: Wiki-over-RAG, Agent Workflows, and Unsolved Multi-Repo Problems
Robin Tegg's April 2026 AI Retrospective covers advancements in persistent knowledge bases, specifically exploring wiki-over-Retrieval-Augmented Generation (RAG) inspired by Andrej Karpathy's work. The review also inclu…
-
Karpathy shifts to agent-driven coding, documents 57,000 failures in CLAUDE.md
Andrej Karpathy shared a markdown file that details his transition to an agent-driven coding workflow, shifting from 80% manual to 80% automated coding within weeks. He documented various failure patterns encountered du…
-
New benchmark 'Prosa' evaluates LLMs on Brazilian Portuguese chats
Researchers have introduced Prosa, a new benchmark designed to evaluate Large Language Models (LLMs) using real user conversations in Brazilian Portuguese. This benchmark utilizes a rubric-based scoring system with mult…
-
Porting microgpt to Futhark, Part I
The author details their experience porting Andrej Karpathy's microgpt, a concise Python implementation of a GPT-2-like neural network, to the data-parallel language Futhark. The goal was to improve scalability beyond P…
-
Mistral’s Model Lets You Vibe Long-Running Code in the Cloud
Mistral AI has released Mistral Medium 3.5, a new 128 billion parameter model designed for extended coding tasks with a 256K context window. This model powers new remote coding agents within Mistral's Vibe platform, ena…
-
Andrej Karpathy: LLMs are now a programmable layer for digital work
Andrej Karpathy discussed a significant shift in programming, where Large Language Models (LLMs) are evolving beyond simple chatbots to become a new programmable layer for digital tasks. He highlighted December 2025 as …
-
Karpathy's Auto-Architecture tool hunts vulnerabilities on GitHub
A humorous post on Mastodon discusses the concept of "Auto-Architecture" in the context of AI code generation, referencing Andrej Karpathy's work and GitHub's code. The author uses the metaphor of a "vulnerability roule…
-
AI research loop optimizes CPU architecture, boosting performance by 92%
An autonomous research loop, inspired by Andrej Karpathy's work, was adapted to optimize a CPU's microarchitecture. The system proposed, implemented, and evaluated hypotheses for a SystemVerilog CPU core, achieving sign…
-
AI models see tool-calling improvements and bug fixes
A new tool has been developed that addresses a need identified by Andrej Karpathy, with its creation reportedly taking only 48 hours. Separately, a bug affecting DeepSeek V4's output in the SGLang open-source inference …
-
Google launches Gemini 3.5 Flash, Omni, and agent stack
Google has launched Gemini 3.5 Flash, a new model designed for agentic workflows and coding tasks, available immediately across its consumer and developer platforms. This release also introduces Gemini Omni for multimod…
-
Interactive guide explains how large language models like ChatGPT are built
A new interactive visual guide, based on Andrej Karpathy's lecture, explains the intricate process of building large language models. It details the journey from collecting vast amounts of internet text to the final sta…
-
Together AI kernels team optimizes GPUs with FlashAttention
The Together AI kernels team, including researchers Dan Fu and Tri Dao, developed FlashAttention, a software layer that significantly optimizes GPU performance for AI models. This breakthrough, achieved by applying data…
-
Data scientists' core skills are essential for AI harness engineering and evaluation
The role of data scientists is evolving with the rise of large language models, shifting from direct model training to a focus on the "harness" that guides AI systems. While foundation model APIs reduce the need for tra…
-
AI coding agents mature, sparking productivity panic and new tools
The AI development landscape has shifted dramatically, with coding agents now capable of sustained, long-horizon tasks, a change noted by Andrej Karpathy since December 2025. This has led to new products like Perplexity…
-
Sequoia Capital backs Flapping Airplanes' data-efficient AI research lab
Venture capital firm Sequoia Capital has invested in Flapping Airplanes, a startup aiming to develop data-efficient AI models. The company, founded by brothers Ben and Asher Spector along with Aidan Smith, focuses on at…