Before diving into code, I wanted to sketch out a rough learning path. This isn't a curriculum — just the topics I think I need to understand, in roughly the order I'll tackle them.

Phase 1 — Foundations

• Python refresher (NumPy, pandas basics)
• Linear algebra intuition (vectors, matrices — not proofs)
• How neural networks work at a high level
• Training vs inference

Phase 2 — Large Language Models

• Transformers and attention (the "what" before the "how")
• Tokenization and embeddings
• Prompt engineering basics
• Fine-tuning vs RAG vs agents

Phase 3 — Building Things

• Calling APIs (OpenAI, Anthropic, local models via Ollama)
• Structured outputs with schemas
• Simple agent loops
• Deploying a small AI-powered app

How I'll document it

Each phase gets blog posts as I go. Some will be short notes, others will be walkthroughs of projects I build.

I'll also keep the resources page updated with links I actually use — courses, docs, tools, and communities.

If this roadmap changes (it will), I'll write about that too. Learning in public means being honest about what you don't know yet.