Dockerfile

FROM python:3.10-slim

WORKDIR /app

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

COPY . .

# Hugging Face Spaces uses port 7860
EXPOSE 7860

ENV FLASK_APP=backend/main.py
ENV PYTHONPATH=/app/backend
ENV HF_HOME=/app/.cache/huggingface
ENV PYTHONUNBUFFERED=1

# Non-root user (HF Spaces security requirement)
RUN useradd -m -u 1000 user && \
    mkdir -p /app/.cache/huggingface && \
    chown -R user:user /app

USER user

ENV HOME=/home/user \
    PATH=/home/user/.local/bin:$PATH

# 1 worker (to maximize CPU/RAM allocation for GGUF), 300s timeout (for slower CPU inference)
CMD gunicorn --workers=1 --timeout=300 --bind=0.0.0.0:${PORT:-7860} backend.main:app