umbertogriffo/Predictive-Maintenance-using-LSTM ? reverse-engineered prompt

Build me a Python project that uses an LSTM model to do predictive maintenance on aircraft engine sensor data.

I want it to take the training and test files from the dataset folder, learn from multiple sensor readings over time, and then do two things: predict how many cycles are left before an engine fails, and also predict whether an engine will fail within a chosen future window.

Please include the full workflow, from loading and preparing the time series data to training, evaluating, and saving the model. I’d also like clear plots or saved outputs that show the loss during training, the prediction quality, and a comparison between actual and predicted results.

If it helps, keep it easy to run in Colab too, and use the same kind of setup this kind of Keras and TensorFlow example would normally use. If you need to check current docs online for anything, go ahead.