Understand AI deeply.
Ship systems that work.
Frontier model deep-dives · AI agent engineering · Cloud architecture & data systems — by zhuermu
Recent
54 entries- 2026-07-20 AI 0-1. Messi's World Cup Is Over. No Models This Time — Just a Proper Goodbye
- 2026-07-18 TOOLS Don't Just Buy Your Engineers AI Tools: The Complete Playbook for Becoming an AI-Native Company
- 2026-07-18 DEEP The Mysterious 400 in Claude Multi-Turn Chats: Debugging a Thinking Signature Corruption
- 2026-07-17 AI Inside Kimi K3's Two Architectural Pillars: KDA and AttnRes, Explained From Zero
- 2026-07-17 AI Kimi K3 Is Here: 2.8 Trillion Parameters, Fully Open Source — This Time It's Different
Topics
AI & Agent
AI agents, LLMs, and autonomous systems
Video Streaming
Codecs, streaming protocols, DRM, and VOD systems
Big Data
Data lakes, query engines, and ML platforms on AWS
Cloud Architecture
AWS, infrastructure, and scalable systems
Coding Tools
IDEs, developer tools, and productivity
Tech Deep Dive
Source code analysis and internals
Latest Posts
View all →
0-1. Messi's World Cup Is Over. No Models This Time — Just a Proper Goodbye
Argentina 0-1 Spain: Ferran Torres' 106th-minute goal ended Messi's sixth and final World Cup. Congratulations to Spain — and a farewell to Leo: the tears after the Egypt comeback, the letter a 15-year-old Enzo Fernández wrote him in 2016, six tournaments across 20 years, and why he was always a human being, never a god.
Don't Just Buy Your Engineers AI Tools: The Complete Playbook for Becoming an AI-Native Company
Why companies that bought every AI coding tool still aren't shipping faster — and the four-layer playbook that actually works: Loop Engineering for individuals, Harness Engineering for teams, AI-DLC 2.0 and AI-PLC for process, and a full organizational redesign of roles, KPIs, policies, and middle management.
The Mysterious 400 in Claude Multi-Turn Chats: Debugging a Thinking Signature Corruption
A customer's Claude extended-thinking conversations kept failing with 400 Invalid signature — but only on some gateway channels. A full postmortem: what the signature mechanism is, how a proxy's JSON re-serialization silently corrupts base64, five controlled experiments that pinned it down, and a fix checklist for every LLM gateway developer.
Inside Kimi K3's Two Architectural Pillars: KDA and AttnRes, Explained From Zero
Beginner-friendly, analogy-driven walkthrough: KDA turns attention's 'open-book exam' into 'one page of smart notes' — 75% less KV-cache memory, 6.3x faster decoding at 1M tokens. AttnRes turns the residual stream's 'running ledger' into 'a notebook with an index' — 25% better training efficiency at under 2% overhead. By the end you'll know exactly how 2.8T parameters stay standing.
Kimi K3 Is Here: 2.8 Trillion Parameters, Fully Open Source — This Time It's Different
Moonshot AI releases Kimi K3: 2.8 trillion parameters, the world's first open-source 3T-class model, 1M-token context, and native multimodality. A deep dive into the KDA and AttnRes architecture innovations, its #4 global ranking on Artificial Analysis, pricing that matches Claude Sonnet 5, and the four long-term ways an open frontier model reshapes the industry.
On the Eve of the Final, I Had fable 5 Run Another Million Simulations: Messi Kicked More Than Half the Gap Away
All six predictions from the last article hit: Spain 2-0 France, Argentina 2-1 England, and the 41.8%-probability Spain–Argentina final came true. Before the final I fed the semifinal data back to fable 5 and re-ran a million simulations: Spain's win probability dropped from 54.1% to 51.8%, Argentina's rose from 45.9% to 48.2% — Messi, leading the tournament scoring chart, took back the points the model had docked for his age. Plus the story of a photo taken 19 years ago: Messi and baby Yamal.