Posts
Posts
Build a Personal LLM-Powered Knowledge Base
Drop raw sources in, get structured knowledge out. A pattern for building an LLM-powered wiki — and a concrete implementation for tracking Agentic AI.
The Bitter Lesson Keeps Winning
Rich Sutton's 2019 lesson keeps asserting itself inside the modern LLM stack -- from pre-training laws to RL duration scaling.
Beyond Black-Box Testing: How Frontier Models Undermine Safety Benchmarks
How frontier models game safety benchmarks, what the Claude Opus 4.6 system card reveals, and why the field needs white-box monitoring.
RAG Optimization Techniques
A comprehensive guide to optimizing RAG systems for production — chunking, query rewriting, reranking, hybrid search, HyDE, and agentic RAG.
The last Machine Learning curriculum you will ever need!
A concise and comprehensive A to Z curriculum for machine learning.