Skip to content

Running Labs in VS Code with the Course Container

Use Visual Studio Code on your host while the Python environment, dependencies, and lab files run inside the Docker container. Edits save to your local disk via the bind mount—no package installs on the host required.

There are two supported workflows:

Workflow Best when…
A. Dev Containers (“Reopen in Container”) You want the full IDE (terminal, debugger, tests, notebooks) inside the course image
B. Attach to running Compose You already started docker compose up for JupyterLab and want VS Code alongside it

Prerequisites

  1. Docker Desktop (Windows / macOS) or Docker Engine + Compose (Linux)
  2. Visual Studio Code
  3. VS Code extensions:
  4. Dev Containers (ms-vscode-remote.remote-containers)
  5. Python (ms-python.python)
  6. Jupyter (ms-toolsai.jupyter) — for .ipynb support

On Windows, keep the repo on the Linux filesystem if you use WSL2 (e.g. \\wsl$\Ubuntu\home\…) for better bind-mount performance.

This uses the repo’s .devcontainer/ config and the same Dockerfile as the Jupyter stack.

Steps

  1. Clone or unzip the course repository and open the repository root in VS Code:
code accurate_secure_rag_systems

  1. When prompted “Reopen in Container”, accept it.
    Or: Command Palette (Ctrl+Shift+P / Cmd+Shift+P) → Dev Containers: Reopen in Container.

  2. VS Code builds the image (first time only), then reloads the window connected as user fischer3.

  3. Open a lab notebook, e.g.:

labs/01-chunking/notebooks/lab-1.1-chunking.ipynb

  1. Select the kernel RAG Course (Python 3.11) or the default Python 3.11 interpreter inside the container.

  2. Run cells with Shift+Enter.

Integrated terminal

Terminal → New Terminal opens a shell inside the container:

whoami                    # fischer3
pwd                       # /home/fischer3/course  (or workspace folder)
cd labs/01-chunking && pytest tests/ -v

Stop / rebuild

  • Command Palette → Dev Containers: Reopen Folder Locally (leave the container)
  • Dev Containers: Rebuild Container after changes to docker/requirements.txt or the Dockerfile

Workflow B — Attach to a running docker compose container

Use this if JupyterLab is already up at http://localhost:8888 and you also want VS Code.

Steps

  1. From the repo root on the host:
docker compose up --build -d

  1. In VS Code: Command Palette → Dev Containers: Attach to Running Container…
    Select rag-course-jupyter.

  2. When the new window opens, choose Open Folder → /home/fischer3/course.

  3. Install the Python and Jupyter extensions in the container if VS Code prompts you (they run on the remote side).

  4. Open notebooks under labs/…/notebooks/ and select the container’s Python 3.11 kernel.

You can use JupyterLab in the browser and VS Code at the same time; both see the same bind-mounted files.

Detach

Close the attached VS Code window. Stop Compose when finished:

docker compose down

Where files live

Location Meaning
Host: your clone of the repo Source of truth on disk
Container: /home/fischer3/course Same files (bind mount)
Container: /home/fischer3/.jupyter Jupyter config (Docker volume)

Saving a notebook in VS Code writes through to the host immediately.

Running tests from VS Code

Option 1 — terminal

cd labs/03-skills && pytest tests/ -v
cd ../04-evaluation && pytest tests/ -v

Option 2 — Testing UI

  1. Install/enable the Python extension in the container.
  2. Command Palette → Python: Configure Tests → pytest → select the lab’s tests folder (or run from each lab directory).

Optional: live GCP from VS Code

Notebooks are offline-first. For Vertex AI / BigQuery paths:

  1. On the host: gcloud auth application-default login
  2. Uncomment the gcloud config volume in docker-compose.yml (or add the same mount under mounts in .devcontainer/devcontainer.json)
  3. Set GOOGLE_CLOUD_PROJECT in the container environment
  4. Rebuild / restart the container

Troubleshooting

Symptom Fix
“Reopen in Container” missing Install the Dev Containers extension
Permission errors on .jupyter docker compose down -v then rebuild / reopen
Wrong Python interpreter Status bar → select /usr/local/bin/python (container)
Notebook kernel won’t start Kernel picker → Python 3.11 / RAG Course (Python 3.11)
Windows path / performance issues Clone the repo inside WSL2 and open it from there
Extensions missing after attach Install Python + Jupyter into container when prompted
Port 8888 in use Only needed for browser JupyterLab; pure Dev Containers does not require publishing 8888