RoboRana

My RoboRana Internship 🐸 — August 2019

Let me begin by briefly introducing myself: I am Edouard Vindevogel, an electrical engineer graduate currently based in Leuven. In this blogpost, I happily take you on a journey throughout my internship experience at RoboRana, where I got the chance to make RPA robots even smarter. Are you wondering how I did this? Keep on reading and learn all about my personal working method, tips and final results!

Discovering the world of Intelligent Automation Because of my great interest in process automation, I was looking for an internship in a similar branche. During a visit at a startup fair in Leuven, I came across several companies involved in the business, including RoboRana. I received an interesting introduction from this young company specialized in Robotic Process Automation (RPA), and after an intense conversation with two passionate RoboRana - RPA engineers, they had my full attention. A few days later, I contacted RoboRana for a chat with the managing partner, Mathias Fransen. After our first encounter, we determined the content of my internship and there I was, ready for the next step!

Making software bots smarter 🤖 RoboRana is always challenging the boundaries of RPA and highly invests in research and development on the so called principle of 'Intelligent Automation'. The integration of RPA and cognitive technologies based on artificial intelligence strengthens and expands the capabilities of RPA itself. Therefore, the main goal of my internship was all about making the software robots at RoboRana smarter. You want specifics? Well, they wanted to know more about the possibilities of Microsoft's Azure services and how they could use this platform to take their current RPA processes to the next level.

The results of machine learning models In a few days, I fully mastered the tools RoboRana uses to make RPA robots. By combining these tools with Microsoft Azure, things like image interpretation, optical character recognition, and text analysis are suddenly possible. I also made a tutorial for my colleagues, including the specific ways of how to combine these services, so they could expand their RPA bots with the,newly learned skills.

Knowing how to use those skills and actually using them in a business case are two very different things. That's why I made a short demo in which I demonstrated the power of combining RPA with NLP and machine learning. (You can also read all about this topic in our previous blogpost!) More specifically, the demo was about the classification of emails, each based on a different topic. I used the Azure platform to create and train a model for this specific assignment.

Creating a machine learning process requires a lot of trained data, so I decided to create a dataset with all sorts of emails from each class. The most suitable model for the training was selected after I did some extensive research in which several models were trained and compared to one another. Eventually, I chose a model that uses a multi-class decision forest with a keyword extractor.

The results of the classification model were then used to forward an email to a specific email address linked on the subject. This model can be used in many domains to help automate business processes of all kinds!

The results are promising, which makes me very optimistic about the future of intelligent automation. As icing on the cake, I had the opportunity to finish my 4-week internship with a presentation to share my accumulated knowledge. 😃

A bright Intelligent Automation future Working at RoboRana has been such a great experience! I learned a lot and got to know a lot of new, cool people! It's a satisfying feeling to have worked out something of great value for RoboRana during my internship. I really look forward to what the future of intelligent automation brings and I can't wait to start my career in this world of automation and software bots!

Do you have any questions after reading this blog post or are you eager to start your internship at RoboRana? Be sure to let us know! 🐸

chevron-up