Developing Impacter: an LLM based proof-of-concept

Published at 2024-02-08 by Ana Rodrigues

After 6 months, Max's internship as come to an end. In this blog post he writes about his experience at Impacter.

On the 4th of September, I started my internship at IDfuse with the goal to improve Impacter by generating suggestions for comments in Microsoft Word with the use of Large Language Models. Now, five months later, with the help of Tijmen Altena as my supervisor, I've created a proof-of-concept of the Impacter update. This addition to Impacter allows the user to generate suggestions for each comment written in Word, enhancing both Impacter's and the user's comments! I am very happy with the final results. I had a lot of fun during this internship and learned a lot. Going into this internship, I expected to learn a lot about programming skills, but this wasn't entirely the case. During my internship, I learned so much about myself and picked up a lot of professional skills.

I would like to thank the whole team at IDfuse for giving me this awesome opportunity and teaching me so much. I would like to give special thanks to Tijmen for guiding me through this project. Special thanks also to UMC Utrecht and the University of Utrecht for helping with the development of the Large Language Model implementation by giving feedback; without them, the project would not have been possible. During the demonstrations of this proof-of-concept to UMC Utrecht and the University of Utrecht, they told me that they see a lot of potential in the generated suggestions, and when finished, this implementation will greatly help with the writing of grant proposals. I find it awesome that I was able to create something that will help with their research.

Many thanks to everyone!

Max Visscher