Jerry

Is an LLM just a database… or a gloriously chaotic, alien one? It compresses the world’s knowledge into parameters only the model itself knows how to query — and every answer subtly shifts like a living, unpredictable mind. #LLM #Database

Perfect timing for this lecture - Memory Hierarchy and Caches youtu.be/l2_TNDtrafI

The 6-part series from Georgia Tech has been a go-to resource for learning high-performance computer architecture.

The biggest AI opportunity isn’t building brand-new apps from scratch. It’s unlocking the massive hidden value inside all the powerful but “dumb” software that already exists. #AgentEngineer #SoftwareIntelligence #AI

• Understand the user’s real intent • Discover relevant features users don’t even know about • Automatically combine workflows • Guide users smoothly to the outcome they need

I officially define a new role: Agent Engineer. Most software is already powerful. The real problem is that users can’t truly access that power. An Agent Engineer embeds intelligent agents into existing software, transforming it so the system can:

Stopped by @mustafap0ly desk today..

Keep this in your C++ toolbox - one of the finest and updated regularly. Optimizing Software in C++ by Agner Fog agner.org/optimize/optim…

Polymarket 支持用户填写 KYC/KYB 表单，称可以获取同机房低延迟接入资格。

教育 SaaS 这下玩出花了。 美国主流学习平台 Canvas 周四被勒索组织 ShinyHunters 攻破下线，正赶上全美期末考周。伊利诺伊大学期末考延期、加州大学系统全线补救。 更离谱的操作在后面。Canvas 登录页直接挂出黑客的赎金信，黑客在上面公开点名母公司 Instructure 拒绝交赎金，让学校自己跟黑客谈。SaaS 厂商把客户的赎金谈判甩给客户。 数据涉及 2.75 亿人、8800 所学校。同一个 ShinyHunters 2024 偷过 Snowflake 客户库，去年 PowerSchool 6000 万学生数据也丢过。

someone analyzed all 5000+ accepted papers at ICLR 2026, and it's a good signal who's pushing the research of AI: > China has surpassed the US with 43.7% of the papers > Europe's contribution is surprisingly small (5.3% including UK)

This is the rarest piece of HFT content on the internet - Robert Almgren walking through a real execution trade tick by tick He's the co-founder of Quantitative Brokers and the man who wrote the Almgren-Chriss model - the standard execution algorithm used at Goldman, Citadel, Two Sigma In this lecture he shows a live order from May 14 2018, every fill, every limit, every market reaction across 2 minutes and 40 seconds The article above is that exact same thinking applied to Polymarket - measure your edge in bits before you place a single trade Bookmark it & give it a watch tonight ↓

Most people don’t think with their head — they think with their ass. If someone bases their entire future direction on their past experiences and current seat, that’s the classic ‘ass determines head.’ They’re not choosing freely — their history is choosing for them.

We implemented @karpathy 's MicroGPT fully on FPGA fabric. No GPU. No PyTorch. No CPU inference loop. Just a transformer burned into hardware, generating 50,000+ tokens/sec. The model is small, but the idea is not: inference does not have to live only in software 👇

See how "g" supports directly reading the content of the specified path of the local file.

I created a terminal agent named "g", which allows for the quickest use of terminal to initiate sessions or search engines without the need for the cumbersome operation of opening a browser.

In this paper is presented SCENIC, an FPGA-based Smart-NIC with end-to-end system integration, designed to support in-network data processing and enabling full customization. arxiv.org/pdf/2604.15128

《TurboMem: HIGH-PERFORMANCE LOCK-FREE MEMORY POOL WITH TRANSPARENT HUGE PAGE AUTO-MERGING FOR DPDK》 arxiv.org/pdf/2603.18690