Python Developer — AI Children's Book Generator (Pollinations.ai / Gemini)

I have a working Python pipeline that auto-generates children's books ready for Amazon KDP publishing. It uses Google Gemini for text and Pollinations.ai (Stitch) for illustrations.
The pipeline already works end-to-end: story generation, image creation, 300 DPI upscaling, quality checks, PDF assembly, and KDP metadata packaging.
What I need:
1. Interactive review loop — after Stitch generates images, pause and ask the user which pages to regenerate before continuing. Simple CLI: accept all / regen page N / skip page N / save & quit.
2. Stitch validation — Pollinations.ai sometimes returns wrong images (wrong subject, colored instead of B&W, watermark, blank). Add immediate post-download validation + auto-retry loop (max 3 attempts with improved prompt).
3. Cover agent — new module to generate a KDP book cover (Stitch generates it, overlay title text, KDP dimensions 2560x1600px).
4. Basic tests — tests/test_pipeline.py is an empty scaffold. Write unit tests for the 4 main agents.
Tech stack: Python 3.10+, Pillow, fpdf2, Google Gemini API, Pollinations.ai REST API
Important: Pollinations.ai can return unexpected results without raising HTTP errors. I need someone who is not afraid to test edge cases, inspect actual HTTP responses, and build robust fallbacks — not just trust the docs.
Deliverables:
• agents/stitch_agent.py — updated with post-download validation
• agents/review_loop.py — new interactive review module
• pipeline.py — updated to use review loop after step 2
• agents/cover_agent.py — new cover generation module
• tests/test_pipeline.py — unit tests
• Short technical note (1 page) on how you handled Stitch edge cases
Budget / timeline: Fixed price preferred. Estimated 3–5h total. Please quote per deliverable if possible.
Full brief available on request. Codebase shared after NDA-lite agreement (just don't redistribute).