Ben Hooper is energized by applying cutting-edge technology to problems that create solutions that are both better and faster than ever before.

Can you share the most interesting part of your career journey?

Working in data science and different industries, there’s always exposure to new technologies that solve a previously extremely hard problem.

One area that has captured my interest is natural language processing (NLP). There are new solutions to dealing with natural language coming out all the time. It’s a very hard area to have a good solution, but with these new tools, like OpenAI’s GPT-3, what would have taken months of machine training time and coding can be done with a simple API call.

Being able to apply my understanding to address these problems and hopefully have a huge impact can be very rewarding.

I’ve been testing out ways that these new tools can be applied to areas like data extraction, classification, and even novel content generation and have been surprised at how advanced these tools have become.

What excites you most about working with data and AI?

The potential to apply cutting-edge technology to problems and have a solution be 10 or even 100 times better or faster than how it’s currently being addressed. Being able to apply my understanding to address these problems and hopefully have a huge impact can be very rewarding.

Describe an interesting project that you have worked on.

Working with another Deloitte group who helps businesses find tax overpayments, they would work on one client project for months, manually finding areas of spend that they could get tax refunds for. A team of five would go through transactions and review documents and invoices, as well as having multiple meetings internally and with the client.

This project was actually what really showed me the potential of even basic machine learning and process automation. 

Using some simple OCR tools and a machine learning model, we were able to create a system that did that work in a few minutes and often recovered millions of dollars more of tax overpayments than the manual process. This project was actually what really showed me the potential of even basic machine learning and process automation.