Computer Engineer, MSc. Researching AI/ML and building from first principles. "What I cannot create, I do not understand." — Richard Feynman
A four-dimension checklist — trigger, structure, steering, pruning — for evaluating whether an AI agent skill actually steers behavior or just burns context-window tokens.
The US walled off frontier AI. Five days later, an open-weight model beat GPT-5.5. The post-training pipeline is where open-source AI compounds its advantage.
How LLMs are built: from internet-scale pre-training through SFT, RLHF, DPO, GRPO, and Constitutional AI.
Sakana Fugu is a learned orchestrator that coordinates frontier LLMs (Claude, GPT, Gemini) to beat each one at its own game. A breakdown of the paper, its two variants, and why orchestration may be AI's third scaling axis.
A first-principles breakdown of why open-source language models are reshaping AI — from cost and transparency to sovereignty, competition, and the now-closed performance gap.