I build and train machine learning and computer vision models in PyTorch, from data cleaning to a model you can call from your app.

My recent work is an image classification ensemble of ResNet50, DenseNet121 and EfficientNet-B0 fused by an attention layer, with augmentation for robustness across datasets.

You get data cleaning and augmentation for your dataset, model training and evaluation with clear metrics, and the trained model with an inference script or a FastAPI endpoint, plus the source code in your repo.

Good fits include image classification, quality and defect checks, and photo or document processing. The price is a starting point for one model on a cleaned dataset. Message me with your dataset and goal and I'll confirm scope.