AI agent architecture, parallel orchestration, skill authoring, knowledge graph systems, and reusable learning frameworks.
This repository documents the AI engineering side of the portfolio — how intelligent agents are designed, how reusable frameworks turn one-off automation into scalable systems, and how learning agents convert any subject into a structured study package. The focus is on architecture and repeatable methodology, not just individual tools.
What's inside:
- 10 AI and learning agents — see the Agent Catalog
- Agent design methodology — two-tier framework for building task and spec-driven agents
- Reusable learning framework — spawns subject-specific study agents from a common base
- Documented frameworks — see the Frameworks collection
- Architecture maps — see the visual index
The engines behind the agents — documented as first-class, reusable artifacts. Includes the Learning Agents Framework (with capability examples and browsable study packages you can study from directly), the Agent Design Framework, the Universal Reasoning Framework, and the Architecture Map Design System.
The methodology for designing new agents — a two-tier framework separating fast task agents (Vibe Coding) from rigorous spec-driven agents, plus the dashboard automation tier-classification system. See the design decisions for the rationale.
Patterns and tooling for formalizing workflows into reusable SKILL.md definitions — frontmatter structure, deterministic vs. agentic step design, triggers, and the test-improve loop.
Knowledge graph systems — entity extraction, relationship modeling, and how structured organizational knowledge feeds agent context.
Parallel orchestration patterns — decomposing complex work into concurrent sub-agents with tracked task groups, schema-driven results, and pipeline composition.
The reusable Learning Agents Framework and the subject-specific agents it produces (CompTIA Security+, SecAI+, AWS AI Practitioner, Agentic AI, and platform coaches). Multi-format output: notes, quizzes, notecards, and assessments. See the design decisions.
Each subfolder is self-contained with its own README. Start with agent-architecture to understand the design methodology, then browse the Agent Catalog to see it applied across ten agents.
Built with Amazon Quick. Agents and frameworks were developed on the Amazon Quick platform and sanitized for public release.
MIT
Part of the SwooshJ-SecAI security and AI engineering portfolio.