I’m developing a deep-learning workflow that turns raw image data into personalized content recommendations. The core of the project is a Convolutional Neural Network, and I want a specialist to take the images I supply, design or fine-tune an appropriate CNN architecture, and train it so that its embeddings can drive a recommendation engine. My stack is flexible, but Python with TensorFlow or PyTorch is preferred because it makes deployment to our existing environment easier. I’ll provide the labelled image dataset and outline the business logic for how the recommendations are surfaced; you focus on data preparation, model design, training, and evaluation. I also need clear metrics (precision, recall or MAP) so I can compare iterations and know when the model is truly production-ready. Deliverables: • Annotated source code and environment file • Trained CNN weights (or a link to the checkpoint) • A short report explaining preprocessing steps, architecture choices, training results, and how to call the model for inference If you’ve already built image-based recommenders or have published work in this area, please mention it when you respond; that experience will be invaluable for shortening my development cycle and improving final accuracy.