My experience

Overview

My main project was an end-to-end data science project creating a deep learning model to perform a classification task. I got to explore the available data and build some interesting features that I eventually created a dataset out of. I also built the pipeline to feed the data into the architecture I had designed using Keras with some custom TensorFlow. I had another project involving using evolutionary algorithms to model how neural networks make predictions that was also very interesting. My key contribution was a front-end tool visualizing image preprocessing and the decision surface of the model.

The projects I was able to work on were super challenging but exciting and I greatly enjoyed them. However, it's a very lonely job. Everyone in my lab was busy all the time and it made it hard to get feedback. Granted, this is during a pandemic and it was great to be able to come into the office to begin with. Code reviews aren't really a thing and you really need to make sure of the work you're putting forward because you are given a solid amount of trust that your results are correct. It's definitely a good job for someone who wants to do heavy research and development and not more of a traditional data scientist role that interfaces with both the code as well as stakeholders.