Knowledge Trees
Prerequisite graphs and knowledge frontiers. Inspectable mastery maps for continuous engineering and systems learning.
The Mechanic: Unlike a linear syllabus, a knowledge tree structures a domain as a directed acyclic graph (per Math Academy's knowledge frontier concept).
Each node specifies a falsifiable demonstration test, exact references, and prerequisite dependencies. You study only at the frontier—items whose prerequisites are mastered.
Published Trees
-
AI Agents
Autonomous agent systems: POMDP formulation, ReAct and Reflexion execution loops, MCP tool calling schemas, 3-tier memory hierarchies, sandboxing, and multi-agent supervisor/worker orchestration.
-
Context Engineering
Designing deterministic token pipelines: prompt caching prefix stability, 4-part structured prompt frameworks, attention degradation mitigation, 3-tier memory hierarchies, and multi-turn chat decontamination.
-
LLM System Design
Architecting production-grade AI systems: model capability spectrum, deterministic context engineering, hybrid vector/graph RAG pipelines, reverse-proxy gateways with tiered fallbacks, and LLM-as-a-judge evals.
-
Online Payments
From typed PAN to merchant settlement. Covers 4-party authorization, CNP fraud liability, EMV 3DS liability shift, Rails multi-provider adapters (ActiveMerchant), and parallel rails (ACH, FedNow, PH InstaPay/PESONet).
-
Bitcoin Protocol
Satoshi Nakamoto's 2008 whitepaper decomposed into assessable protocol mechanics: double-spending, proof-of-work consensus, Merkle pruning, Simplified Payment Verification (SPV), and Gambler's Ruin security proof.