A Query Engine: A 7-Part Deep Dive in Rust
Handwritten parser for clarity teaches query engine internals with working Rust code.
Documentation, Prompts, and Media for the "Decoding the Language Machine" series
CTO with Penn PhD demystifies LLMs through Shannon's 1948 information theory.
Students and developers wanting to understand LLM fundamentals historically
3Blue1Brown · StatQuest · Andrej Karpathy lectures
- Main Site: https://skepticcto.com/ (includes related AI news articles)
- Code/Artifacts: https://github.com/SkepticCTO/decoding_the_language_machine
- YouTube Channel: https://www.youtube.com/@SkepticCTO
I’m a 21-year CTO, Ph.D. in CS (U Penn, 1999 in computer vision and ML), and a PI in the NIST AI Safety Initiative Consortium. I spent a 4-month sabbatical making this because I wanted to demystify how LLMs work through a historical perspective (starting in 1948 with Claude Shannon) and scientific skepticism.
The project is old enough to be fleshed out, but young enough to be able to pivot. Is it useful? What would you like to see? I look forward to questions and feedback.
Handwritten parser for clarity teaches query engine internals with working Rust code.
Clean trivia quiz interface, but Duolingo and Clozemaster already solve this better.
ANML proposes a new protocol for agents, but adoption depends on an unproven agentic web.
Thirteen animated episodes on LSM trees from first principles to production.
The tinvaak.js CLI directly maps whimsical Dovahzul keywords (e.g. gron → let, hah → function, ofan → return) into runnable JavaScript and ships with examples and installable npm scaffolding. It's an enjoyable, well-documented novelty that's useful for teaching transpiler basics or just writing silly Skyrim-flavored scripts — but it's firmly a niche, playful project rather than a tool with broad practical value.
Synthesizes 17 quantum proposals into actionable Bitcoin security advice without the FUD.