Custom AI Image Generation Tool for DTC Incontinence Product Modeling We sell incontinence underwear DTC and need a custom internal tool to generate product model images at scale using AI. Off-the-shelf AI tools don't work for us — they change the product details (wrong seam placement, wrong padding shape, wrong fabric look). So we need a purpose-built system that locks in product accuracy through structured product profiles. What you'll build: A web-based tool where we define a product and a scene direction, kick off a batch, and come back to a gallery of 30+ generated images to pick from. The core is a Gemini Pro image generation engine (not Flash — this is hardcoded) backed by a job queue so batches run in the background. Products are defined through a profile system — fabric type, seam locations, padding description, colorways, fit notes — which the prompt engine reads to assemble accurate generation prompts every time. Users can also save and reuse prompt templates through a prompt library. There are 3 generation modes: starting from scratch with just a product profile, adapting poses from an existing product image to a new product, and generating from a saved prompt template. Tech stack: Node.js or Python backend, BullMQ + Redis for the job queue, PostgreSQL for profiles and job records, S3 for image storage, React frontend. Scope: ~20 hours This is a greenfield build — there's a prior version that exists for reference but this is a clean rebuild with a different architecture.