The A1 → A5 journey library
15 journeys — each is a 12-chapter mentored deep-dive following the same template as the URL Shortener flagship. Every technical claim is cited (RFC / paper / vendor docs / engineering blog). Same mentor voice.
Start with the A1 flagship — first LLM API call
Model selection (GPT-4o vs Claude 3.5 Sonnet vs Gemini 2.5 Pro — pricing + capability head-to-head), token math with tiktoken, streaming, cost dashboard, rate limit realities. Same rigor as URL Shortener.
API Caller
Zero to first LLM API call
A1 flagship — model choice, tokens, streaming, cost
Ship your first LLM-powered endpoint in an afternoon — with the same rigor as the URL Shortener journey.
From A1 to A5 — the meta-journey
How to skill up in AI without wasting years
The learning path a senior AI architect wishes they'd followed — decision by decision.
Prompt Engineer
Prompt engineering fundamentals
Zero-shot, few-shot, CoT, and when each wins
A senior prompt engineer walks you from ad-hoc prompts to reproducible, versioned, evaluated prompts.
Structured outputs & function schemas
JSON mode, tool schemas, Pydantic/Zod validation
How production systems get LLMs to output exactly the shape you need — every time.
RAG Builder
Build your first RAG system
Chunk, embed, retrieve, ground — end to end
The canonical A3 journey — from a folder of PDFs to a grounded Q&A endpoint with citations.
Hybrid search + reranking
Dense + sparse + BM25 + cross-encoder
When vanilla vector search stops working — and how to build hybrid retrieval that actually beats it.
Agent Engineer
Build your first agent (ReAct)
Reasoning + acting + observation loop
From API caller to agent — build a ReAct loop with tools that actually ships to prod.
Build your first MCP server
Model Context Protocol from scratch
The new interoperability standard for LLMs and tools — build a working MCP server end to end.
Agentic loop deep-dive
ReAct → Reflexion → Tree-of-Thought → Multi-agent
The full spectrum of agent loop architectures — when each wins, when each fails.
AI Architect
Ship an AI feature to production
Evals, guardrails, observability, cost, rollback
The production checklist a senior AI architect uses before shipping — every item defended.
Defend against prompt injection
The OWASP AI Top 10, applied
The attack surface no one taught you — with real CVEs, defense-in-depth, and testing.
LLM evals that matter
Beyond BLEU — LLM-as-judge, golden datasets, human-in-the-loop
How production teams actually evaluate LLM output — and what to do when metrics disagree.
Cost optimization for LLM apps
Prompt caching, batching, model routing, distillation
How to cut your LLM bill 10x without losing quality — with the math to defend every choice.
Multi-agent systems
Orchestrator + workers, hierarchical, market-based
When one agent isn't enough — and the design patterns that keep multi-agent systems from imploding.
Fine-tuning vs RAG — the decision
When to fine-tune, when to RAG, when to do both
The single biggest architectural fork in AI — with the cost, quality, and complexity math.