Dockerfile

FROM python:3.9-slim

# System dependencies
RUN apt-get update && apt-get install -y \
    git \
    gcc \
    g++ \
    curl \
    libopenblas-dev \
    python3-dev \
    && rm -rf /var/lib/apt/lists/*

# Create cache directory
RUN mkdir -p /app/cache && chmod 777 /app/cache

WORKDIR /app

# Environment variables
ENV TRANSFORMERS_CACHE=/app/cache \
    HF_HOME=/app/cache \
    PYTHONWARNINGS="ignore" \
    NPY_NO_DEPRECATED_API=1 \
    XDG_CACHE_HOME=/app/cache \
    PRELOAD_MODEL=false \
    TOKENIZERS_PARALLELISM=false

COPY requirements.txt .

# Install Python deps in order (numpy first)
RUN pip install --no-cache-dir --upgrade pip && \
    pip install --no-cache-dir numpy==1.26.4 && \
    pip install --no-cache-dir -r requirements.txt

COPY app.py .

EXPOSE 5000

# Healthcheck
HEALTHCHECK --interval=30s --timeout=30s --start-period=60s --retries=3 \
  CMD curl -f http://localhost:5000/health || exit 1

# Use a single worker to reduce memory pressure
CMD ["gunicorn", "app:app", "--workers=1", "--threads=1", "--timeout=300", "--bind=0.0.0.0:5000"]