Dockerfile

FROM python:3.10-slim

# Install system dependencies
RUN apt-get update && apt-get install -y \
    build-essential \
    python3-dev \
    curl \
    && rm -rf /var/lib/apt/lists/*

# Set up the Hugging Face standard user
RUN useradd -m -u 1000 user
USER user
ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH
WORKDIR $HOME/app

# CRITICAL FIX: Tell pip NOT to use a cached build, and enforce a clean wheel
ENV PIP_PREFER_BINARY=1

# Install llama-cpp-python using pre-compiled wheels for CPU
RUN pip install --no-cache-dir llama-cpp-python --extra-index-url https://abetlen.github.io/llama-cpp-python/whl/cpu

# Copy requirements and install the remaining packages (FastAPI, etc.)
COPY --chown=user requirements.txt .
RUN pip install --no-cache-dir -r requirements.txt

COPY --chown=user app.py .

EXPOSE 7860
CMD ["python", "app.py"]