During my internship (combined with my Master thesis), I developed a data product to monitor FDA submissions for specific topics. The automatized pipeline included downloading the latest FDA pdf's, extracting their contents, and retrieving the documents mentioning pre-defined topics of interest. The results were then displayed on a Google Data Studio dashboard.In order to improve the topic extraction, I developed various natural language processing and machine learning models. All sentences were encoded using BERT, then classified as relevant/non-relevant using a K-Nearest Neighbours model. This allowed me to find other occurrences of the topics of interest.I also created a R-Shiny application to allow our users to search the full text of the downloaded FDA documents. The documents were indexed in an ElasticSearch node, as it allows fast text queries and efficient storage.

The internship consisted of backend software development in a Blockchain environment. My main tasks included transactions database optimizations and code documentation automatization. The main programming language was Java, but I also learned the basics of Kotlin. We relied on GitLab as a version control system. I learned to use Docker, as well as GitLab CI/CD pipelines for code testing.

Assistant for the EPFL Algorithms course. Grading and correction of exercises. Answering student questions and providing further explanations during weekly exercise sessions.