Each project is different; some may require delving deep into the data, some may need Dimensionality Reduction, others may need heavy feature engineering... Let's discuss your project and find the best way to move with it.

I will handle missing data, create Training/Evaluation/Testing sets, stratified if needed, perform transformations on all non-numeric variables so the algorithms can work with them, perform statistical transformations on the dependent variable (if numeric), and more.

After the data is ready, I will create on PyTorch a Deep Neural Network (Feed-Forward, Convolutional or Recurrent, depending on the problem) configured to take advantage of a GPU if available, or else go to the CPU.

I will export the model, and expose it either publically or locally through a RESTful API, if needed.