A living knowledge system for AI Engineers who build, not just collect.

ROAR is my personal knowledge base and public portfolio — a structured, curated repository of everything that matters in AI Engineering, GenAI Solution Architecture, and beyond.

It is not an awesome-list. It is not a bookmark dump.

Every entry has three layers:

• Why it matters — relevance to real systems and real decisions
• Key insight — the one thing worth remembering
• How I'd use it — concrete application: RAG pipeline, agentic workflow, enterprise architecture, security review, project delivery

New entries land daily. The system is designed to scale to thousands of entries without becoming unreadable.

Primarily: me — Przemo, AI Engineer / GenAI Solution Architect.

Secondarily: anyone building serious AI systems who wants curated signal instead of noise.

The repo concentrates on:

• LLM applications — prompting, fine-tuning, evaluation
• RAG pipelines — chunking strategies, retrieval patterns, vector databases
• Agentic workflows — LangGraph, multi-agent orchestration, tool use
• GenAI architecture — modular system design, POC→MVP patterns, enterprise constraints
• Python AI ecosystem — frameworks, tooling, local model runners
• Cybersecurity for AI systems — offensive methods, defensive practices, OWASP for LLMs, secure agent design
• Project & delivery practices — Agile, ITIL, PRINCE2, PMI in the context of large-scale GenAI initiatives

Use GitHub's built-in search or browse folders directly.

Tips:

• Search by tag (e.g. rag, agents, langgraph) across all .md files
• Start with curated/best-of.md for the highest-signal entries
• Read learning-notes/2026/ for synthesized patterns and architectural decisions — this is where links become understanding

Full structure below (folders only — entries are individual dated .md files)

ROAR/
├── inbox/                     # Quick capture — categorized later
│
├── repos/
│   ├── ai/
│   │   ├── llm-and-rag/
│   │   ├── agents-and-orchestration/
│   │   ├── frameworks-and-tooling/
│   │   └── architecture-and-patterns/
│   ├── devops/
│   │   ├── infra-and-cloud/
│   │   ├── ci-cd-and-automation/
│   │   └── monitoring/
│   ├── cybersecurity/
│   │   ├── offensive/
│   │   ├── defensive/
│   │   └── frameworks/
│   ├── project-management/
│   │   ├── agile-and-scrum/
│   │   ├── itil/
│   │   └── prince2-and-pmi/
│   └── cool-repos/
│       ├── productivity-and-automation/
│       ├── data-and-documents/
│       └── misc/
│
├── youtube/
│   ├── ai-and-ml/
│   ├── industry-and-news/
│   └── engineering-and-tools/
│
├── prompts/
│   ├── ai-engineering/
│   ├── coding-and-review/
│   ├── meta-prompts/
│   ├── productivity/
│   ├── photography/
│   └── misc/
│
├── experiments/               # Personal scratchpad — mini RAGs, POCs
│
├── curated/
│   └── best-of.md             # ≤20 entries — only what I'd recommend
│                              # to any senior AI Engineer in a job interview
│
└── learning-notes/
    └── 2026/                  # Weekly synthesis: patterns, decisions, takeaways

Every file follows this format:

---
date: 2026-04-20
tags: [rag, langgraph, production]
importance: high
status: applied
source: https://...
author: @handle
---

# Tool / Repo / Resource Name

## Why it matters
Why this is relevant to an AI Engineer or Solution Architect.

## Key insight
The one concrete thing worth remembering.

## How I'd use it
Specific context: RAG pipeline, agentic workflow, enterprise design,
security review, modular architecture — whatever applies.

## Related
- Links to related entries in this repo

Every entry includes YAML frontmatter + Why it matters + Key insight + How I'd use it + Related (optional).

status tracks real lifecycle: to-review → reviewed → applied → discarded

importance reflects relevance to real-world AI systems and architectural decisions: high / medium / low

→ curated/best-of.md

Twenty entries maximum. The bar: would I recommend this to a senior AI Engineer preparing for a Solution Architect interview?

→ learning-notes/2026/

Weekly synthesis. Not a journal — structured reflection:

• What I learned
• Patterns I noticed
• What I want to test next
• Decisions I made (and why)
• One-sentence takeaway for interviews

This is the highest-value part of the repo. This is where links become understanding.

rag llm agents agentic orchestration langgraph langchain vector-db embeddings chunking retrieval fine-tuning evals mlops infra cloud ci-cd monitoring python framework local-model architecture patterns modular enterprise solution-architecture production scalability devops automation productivity prompt meta-prompt cybersecurity pentesting offensive-security defensive-security owasp nist iso27001 agile scrum itil prince2 project-management risk-management

Daily. The green graph is intentional — consistency is a signal.

Built and maintained by Przemo — AI Engineer / GenAI Solution Architect. Opinions are mine. Curation is deliberate.