Study Path

No single course covers the full curriculum. This sequences the best parts of each.

Phase 1 — Foundations (Modules 1–3)

• Karpathy Neural Networks: Zero to Hero — videos 1–4 only; optional if you’re focused on building, not understanding models from scratch
• Anthropic Prompt Engineering — all 9 chapters; the best single resource for Modules 2–3

Phase 2 — Retrieval (Modules 4–5)

• DeepLearning.AI Vector Databases — full course, covers Module 4
• DeepLearning.AI Advanced RAG — full course; focus on chunking and eval patterns, skip LlamaIndex abstractions

Phase 3 — Systems (Modules 6–9)

• AI Agents in LangGraph — full course; covers tool calling, agents, and memory together — skip the standalone memory course, heavy overlap
• MCP with Anthropic — full course; do this after the agents course

Phase 4 — Evaluation (Module 10)

• Intro to LangSmith — full course; pair with RAGAS docs for scoring patterns

Phase 5 — Open-Source and Multimodal (Modules 11–12)

• Hugging Face NLP Course — chapters 1–4; understand transformers, tokenizers, and the Hugging Face ecosystem
• DeepLearning.AI: Building Multimodal Search and RAG — full course; covers multimodal embeddings and retrieval

Phase 6 — Fine-Tuning (Module 13)

• DeepLearning.AI: Finetuning Large Language Models — full course; covers data prep, training, and evaluation
• Hugging Face PEFT docs — for LoRA/QLoRA on open models

Phase 7 — Production (Modules 14–18)

• OWASP Top 10 for LLM Applications — read the full list; this is your safety checklist
• DeepLearning.AI: LLMOps — full course; covers deployment, monitoring, and automation pipelines

Certification

Anthropic Claude Academy