Enabling Artificial Intelligence by quantifying its risks

As Ph.D. candidate, I was responsible of one main project related to autoML and natural language processing and I was collaborating in other projects, more related to computational biology. Particularly, the main project of the thesis was helping users to use machine learning algorithms and data science pipeline without the need of coding. To do this, the first year I had to design and develop a multimodal conversational agent that was helping biologists and clinicians in order to extract and analyze genomic data. The second year, I worked with a more generic purpose audience and I developed a tool composed of a machine translator that converts a natural language sentence into a ML pipeline and an autoML system able to execute it. The third year, I worked with the group at MIT on a event-based sequence data. The objective was a library able to generate automatically prediction tasks and the labels of the data, to filter these prediction tasks using active learning methods and finally to provide an accuracy to the user using autoML.Other projects that I worked on were regarding the development and the analysis of genomic data to extract genes biomarkers for a specific type of cancer and the design and the analysis of the NMTF algorithm for drug repurposing.

I was responsible of organizing and teaching the recitations for Bioinformatics Master's students. Particularly, I was designing Python exercises and I was responsible of explaining how to solve them. I also gave some lectures regarding some Python topics.

As laboratory assistant of the course Foundation of Informatics for electrical engineering students, I had to prepare the exercises and help the students during the execution. Particularly, the topics of the course were explaining to the students how to code in C and in MATLAB.

I was responsible of working on a project on event-based sequence data. The objective was a library able to generate automatically prediction tasks and the labels of the data, to filter these prediction tasks using active learning methods and finally to provide an accuracy to the user using autoML.Finally, I also collaborated with a company on analyzing wind turbines data in order to provide them with an easy pipeline for the faults prediction.

Private tutor of math, physics, english, latin to high school and middle school students.