Skip to content

Setup Guides

Last reviewed: October 2026

Five hands-on guides take you from an empty folder to a working RAG app and a tool-using agent. They are written for engineers preparing for senior, staff and principal AI and platform interviews who want to have built the thing, not just read about it. Each guide has runnable code, the expected output, troubleshooting for common errors, cost and safety notes, and a section on how the topic comes up in interviews.

No API key? You can still do every guide

Each guide offers a local path with Ollama, so it is free and your data stays on your machine. Hosted options (OpenAI, Anthropic, Amazon Bedrock) are shown side by side when you want them.

Suggested path

Work through them in order. Each guide builds on files from the one before.

flowchart LR
    ENV[1. Local environment<br/>uv, Python 3.13, .env] --> LLM[2. LLM access<br/>first model call]
    LLM --> VDB[3. Vector DB<br/>embed + search]
    VDB --> RAG[4. First RAG app<br/>grounded answers]
    RAG --> AGENT[5. First agent<br/>tool-use loop]

Budget about half a day for the whole path. The first two guides take minutes.

  • 1. Local Environment


    Install uv, pin Python 3.13, add dependencies and keep API keys out of git.

    Open guide

  • 2. LLM Access


    A first call with OpenAI, Anthropic, Bedrock or local Ollama, plus a provider-neutral complete() helper.

    Open guide

  • 3. Vector DB Setup


    Embed text and run a similarity search in Chroma, Qdrant, LanceDB or pgvector.

    Open guide

  • 4. First RAG App


    Chunk, index and retrieve your documents, then answer with citations and a safe "I don't know" when the answer isn't there.

    Open guide

  • 5. First Agent


    A hand-written tool-use loop with step limits, then the same agent with LangChain create_agent.

    Open guide

Before you start

  • A terminal and git.
  • Either an API key from OpenAI or Anthropic, AWS access to Bedrock, or a machine that can run a small local model (about 8 GB RAM).
  • You don't need to install Python yourself. Guide 1 installs it with uv.

Keep keys safe

Every guide reads keys from environment variables in a git-ignored .env file. Never commit a key. If one leaks, revoke it straight away.

After the guides