Docker for Beginners: Containerizing My First ML Model

Sample post. A short tutorial-style article showing Dockerfile syntax highlighting and a step-by-step structure. Replace with your own notes whenever you like.

Why Bother with Docker

"Works on my machine" is a real problem the moment a model needs to run somewhere else — a teammate's laptop, a demo server, a CI pipeline. Docker's whole pitch is that the environment ships with the code.

The Minimum Dockerfile

For a Python ML project, this is roughly the smallest Dockerfile that's actually useful:

FROM python:3.11-slim

WORKDIR /app

COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY . .

CMD ["python", "main.py"]

Four real instructions: pick a base image, install dependencies, copy the code, define the run command.

Building and Running

docker build -t my-model .
docker run --rm my-model

--rm cleans up the container after it exits, which is what you want for anything that isn't a long-running service.

The Mistake I Kept Making

Copying the requirements file separately from the rest of the code, before RUN pip install, isn't just style — it's what lets Docker cache that layer. Copy everything at once and every code change forces a full dependency reinstall on the next build.

# Slow: any code change reinstalls every dependency
COPY . .
RUN pip install --no-cache-dir -r requirements.txt

# Fast: dependencies only reinstall when requirements.txt changes
COPY requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt
COPY . .

What I'd Tell a Beginner

You don't need to understand multi-stage builds or image layering theory to get real value from Docker. You need one working Dockerfile, docker build, and docker run. Everything else is an optimization you can learn once the basics are already paying off.