Dockerfile

# ── Mochiva HF Space — CPU inference server ──────────────────────────────────
# Base: Python 3.11 slim (small image, fast startup on HF free tier)
FROM python:3.11-slim

# HF Spaces runs as user 1000 — set up a non-root user
RUN useradd -m -u 1000 mochiva
WORKDIR /app
RUN chown mochiva /app

# ── System dependencies ────────────────────────────────────────────────────
# Only what we strictly need: no CUDA, no build tools for heavy packages
RUN apt-get update && apt-get install -y --no-install-recommends \
        git \
        curl \
    && apt-get clean \
    && rm -rf /var/lib/apt/lists/*

# ── Python dependencies ────────────────────────────────────────────────────
COPY requirements.txt .
RUN pip install --no-cache-dir --upgrade pip \
 && pip install --no-cache-dir -r requirements.txt

# ── App code ───────────────────────────────────────────────────────────────
COPY app.py .

# ── HF Spaces metadata ────────────────────────────────────────────────────
# Port 7860 is the standard HF Space port
EXPOSE 7860

# ── Run as non-root ────────────────────────────────────────────────────────
USER mochiva

# ── Startup ────────────────────────────────────────────────────────────────
# --workers 1: model is loaded once in the main process; threading handles concurrency
# --timeout-keep-alive 30: keep SSE connections alive
CMD ["uvicorn", "app:app", \
     "--host", "0.0.0.0", \
     "--port", "7860", \
     "--workers", "1", \
     "--timeout-keep-alive", "30", \
     "--log-level", "info"]