Here is a practical path through this and the earlier courses. The time estimates are rough and assume steady study; they will be shorter if you already know JavaScript or Python.
Suggested order
- Foundations (2 to 4 weeks): programming, machine learning and LLM basics. Start with Generative AI & LLM Fundamentals.
- Prompting and the Claude API (1 to 2 weeks): see Claude API & Tool Use.
- Structured outputs and tool use (1 to 2 weeks).
- Embeddings, vector databases and RAG (2 to 4 weeks).
- Agents and MCP (2 to 4 weeks).
- Evaluation, safety and deployment (2 to 3 weeks).
- A portfolio project (3 to 6 weeks).
Five projects
- Chatbot: a streaming chat app with a system prompt, history, error handling and token tracking.
- Document Q&A with RAG: upload files, chunk, embed, retrieve and answer with citations.
- Tool-using agent: give a model safe tools, validate arguments and evaluate task completion.
- Resume upgrade tool: generate resume text, match it to a job description, return structured resume JSON, give ATS-style feedback and export a clean PDF. Try our ATS checker and job matcher to see what such tools produce.
- MCP assistant: build an MCP server with a few controlled tools and connect it to an AI client.
Choosing a model for your own product
Before switching providers, run the same test set on each candidate and compare resume quality, accuracy, latency and cost. The numbers from your own cases decide, not general rankings.
Habits that make you better
- Build small things often and put them on GitHub.
- Write down what failed and why.
- Read the official documentation, because features change.
- Keep evaluation and safety in every project from day one.
Building something you can show helps with job applications. When you are ready, build your resume with Prime Resume AI and look for roles in our jobs section.