Containerizing and Dockerizing AI Applications

+-------------------------------------------------------------------------------+
|                       MULTI-STAGE DOCKER BUILD FOR AI                         |
+-------------------------------------------------------------------------------+
|  Stage 1 (Builder): Heavy CUDA compiler, build tools, wheel downloads (~4GB)  |
|                                     |                                         |
|                                     v (copy only compiled wheels)             |
|  Stage 2 (Runtime): Slim Python base image, minimal dependencies (~800MB)     |
+-------------------------------------------------------------------------------+

Challenges in AI Containerization

• Giant Image Sizes: PyTorch with CUDA support alone is over 2.5 GB.
• GPU Driver Mismatches: Ensuring host NVIDIA driver is compatible with container CUDA runtime.
• Layer Caching: Preventing costly pip install torch re-runs on every minor code tweak.

Production Multi-Stage Dockerfile

# Stage 1: Build & Dependencies
FROM python:3.11-slim as builder

WORKDIR /build

RUN apt-get update && apt-get install -y --no-install-recommends     build-essential     curl     && rm -rf /var/lib/apt/lists/*

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

# Stage 2: Minimal Runtime
FROM python:3.11-slim as runtime

WORKDIR /app

# Non-root security user
RUN useradd -m -u 1000 appuser

# Copy installed Python packages from builder
COPY --from=builder /root/.local /home/appuser/.local
ENV PATH=/home/appuser/.local/bin:$PATH

COPY . /app
RUN chown -R appuser:appuser /app

USER appuser

EXPOSE 8000

HEALTHCHECK --interval=30s --timeout=5s --start-period=5s --retries=3   CMD curl -f http://localhost:8000/health || exit 1

CMD ["uvicorn", "main:app", "--host", "0.0.0.0", "--port", "8000", "--workers", "2"]