Visual, interactive developer roadmaps — step-by-step guides for building modern AI & application systems, shared as clean, self-contained HTML artifacts. Each one is a single-file, interactive recreation in the spirit of the roadmap-style guides developers already love (à la roadmap.sh) — no build step, just open it in a browser.
Live gallery: https://fahdi.github.io/dev-roadmaps/
| Roadmap | Description |
|---|---|
| AI Application Architecture | Frontend → Agent Orchestration → Knowledge / Models / Tools → Data Layer → Observability & Evals → Deployment |
| How to Build a RAG | Load → Chunk → Embed & Tag → Index → Retrieve → Augment → Generate → Evaluate (and iterate) |
| RAG vs LoRA vs Full Fine-Tuning | Context, adapters, or every weight — pick by the gap: knowledge → RAG, behavior → train |
| Design Your Agent's Control Plane | CLAUDE.md → Rules → Skills → Agents → Hooks — context tells, skills teach, agents delegate, hooks enforce |
| 6 Levels of Agentic AI | Basic → Router → Tool Calling → Multi-agent → Autonomous → Loop Engineering — levels 1–5 are capabilities, level 6 is the craft |
| Context Engineering | The window is a budget: write, select, compress, isolate — and the four failure modes of long context |
| How a Request Travels | DNS → TLS → CDN → Load Balancer → API Gateway → Service → Cache → Database — with the two cache shortcuts |
| How to Evaluate an LLM App | Golden dataset → four evaluators → offline baseline → CI/CD gate → guardrails → online evals — failures become test cases |
index.html gallery landing page (published via GitHub Pages)
artifacts/*.html one self-contained HTML roadmap per file
- Drop a new self-contained
.htmlfile intoartifacts/. - Add a card for it to
index.htmland a row to the table above. - Commit and push — GitHub Pages redeploys automatically.
Built by Fahad Murtaza with Claude.