Computer Vision Engineer for Image App

Looking for a computer vision / generative image engineer to improve object replacement in an existing image-editing app. The main challenge is accurate object scale, placement, and perspective while maintaining realistic lighting and blending on photos front on but also from angles. Experience with OpenCV, masks, inpainting, depth estimation, Gemini, Gpt-image 2, FLUX, Qwen, ControlNet, or Adobe Firefly is preferred. Currently utilising SAM 2,3 Depth anything and a 2 point approach to get rough item dimensions in the photos, then creating a mask overlay and sending to gemini flash or gpt-image 2, it works well until those models sometimes ignore the mask