Senior AI Computer Vision Engineer – Virtual Try-On for Fashion & Saree Catalog Generation

# AI Computer Vision Engineer – Build High-Fidelity Virtual Try-On Workflow for Premium Saree Catalog ## Project Overview We are a premium Indian fashion brand specializing in handcrafted sarees, including Ajrakh, Kalamkari, Bandhani, and hand-embroidered collections. We are looking for an experienced **AI Computer Vision / Generative AI Engineer** to develop a high-fidelity virtual try-on workflow that generates premium ecommerce catalog images by replacing only the saree while preserving the original model and studio photography. This is **not** a prompt engineering or Midjourney project. We require expertise in AI fashion, virtual try-on, diffusion models, and computer vision. --- ## Current Assets We already have: * 2 professional female models * Studio-quality photography * Consistent lighting setup * White seamless backdrop * Professional camera setup * 5 fixed poses per saree * 24 professionally photographed sarees * High-resolution images * High-resolution images of new sarees for garment replacement Target Scale: * 180 sarees * 5 catalog images per saree * Approximately 900 final images --- ## Project Objective Build an AI workflow that: * Uses our existing model photographs as the base. * Replaces only the saree with a new saree design. * Produces photorealistic, premium ecommerce-quality images. The generated images must preserve the original photography while accurately transferring the new saree. --- ## Mandatory Output Requirements The AI workflow must preserve without modification: ### Model * Face identity * Facial features * Skin tone * Hairstyle * Makeup * Body proportions * Hands * Expression ### Photography * Pose * Camera angle * Perspective * Lens characteristics * Lighting * Shadows * Background * Backdrop folds * Floor * Studio environment ### Saree The transferred saree must retain: * Exact color * Exact print * Exact motif placement * Ajrakh / Kalamkari artwork * Border width * Border design * Pallu design * Pleats * Fabric drape * Fabric fall * Texture * Weave appearance * Natural dye appearance * Embroidery * Mirror work * Thread work * Surface details The AI must **not** invent, simplify, recolor, or modify any textile design elements. --- ## Expected Quality The final output should be suitable for: * Ecommerce product catalogs * Website product pages * Marketplace listings * Premium fashion marketing Images should look indistinguishable from a professional studio photoshoot. --- ## Technical Skills Required Candidates must have proven experience with one or more of the following: * ComfyUI * FLUX.1 / FLUX Kontext * IDM-VTON * StableVITON * CatVTON * ControlNet * IP-Adapter / FaceID * LoRA Training * Human Parsing * OpenPose * Segment Anything (SAM) * Diffusion Models * PyTorch * Python * AI Image Inpainting * AI Upscaling Equivalent technologies are welcome if they achieve the required quality. --- ## Deliverables * Complete AI workflow for virtual try-on * Workflow documentation * Installation and setup instructions * Batch processing capability * Prompt/configuration files (if applicable) * Quality testing on our sample images * Optimized workflow for generating approximately 900 catalog images --- ## Proposal Requirements Please include: 1. Similar fashion AI or virtual try-on projects. 2. Portfolio, GitHub, demo, or sample outputs. 3. Technologies you recommend for this project. 4. Your proposed workflow. 5. Estimated timeline. 6. Estimated cloud GPU or hardware requirements. 7. Your approach to preserving garment accuracy, print fidelity, and model identity. Only apply if you have **hands-on experience building AI virtual try-on or fashion computer vision workflows**. Generic AI image generation or prompt engineering experience alone is not sufficient. To confirm you have read this description, begin your proposal with: **AI Fashion Pipeline**