Project Overview: We need an experienced AI developer to build and train a fast, accurate Sky Replacement System with natural blending and sharp masking quality. The system must detect skies, replace them realistically, and enhance overall image quality. Main Requirements: Train the AI model using our image dataset for precise sky detection and highly realistic results (training on our data is mandatory for accuracy). Automatically skip processing if the detected sky area is less than 5% to save server resources. Use advanced blending techniques (Gradient-Domain, Poisson, or Edge-Aware) to achieve natural lighting and smooth edge transitions. Ensure sharp and detailed masking, especially around trees, buildings, people, and hairlines. System must be very fast and fully optimized for server-side deployment (all implementation will be done on our server). Image Optimization: Reduce image size during upload to minimize server load. Use AI Super Resolution / Enhancer (e.g., Real-ESRGAN or SwinIR) to upscale and enhance images when downloading from the web. Deliverables: Fully trained AI sky replacement and super resolution models. Source code, pretrained weights, and complete setup documentation. Skills Needed: PyTorch / TensorFlow / ONNX OpenCV, NumPy, image segmentation and blending Proven experience in AI training for realistic photo enhancement and editing Deadline: The complete project, including training and delivery, must be finished within 2 weeks.