cv
General Information
Full Name | Daniel Scalena |
Date of Birth | 11th September 1999 |
Languages | Italian, English |
Education
-
2023
PhD, Computer Science
University of Milano - Bicocca & University of Groningen
- Interpretability as a tool to make generative models safer, more reliable and less toxic.
-
2023
MSc, Computer Science
University of Milano - Bicocca
- Final grade 110/110 with Honors.
- Thesis "On the explainability of Large Language Models detoxification".
-
2021
BSc, Computer Science
University of Milano - Bicocca
- Final grade 110/110 with Honors.
- Thesis on "Hate Speech detection using NLP techniques on Tik Tok social network".
Experience
-
2023 - Present
Research Intern
University of Groningen, Netherlands
- Study, research and development of RLHF/RLAIF and fine-tune algorithms applied to generative language models to modify their behaviour in generating hate speech, effectively automating and controlling the detoxification process and counter-narrative generation
- Research on the interpretability of the models themselves, analyzing their shift as a result of the post-training procedures adopted.
-
Jun 2022 - Present
Assistant Researcher
University of Milano - Bicocca, Italy
- Research project commissioned by the Italian Ministry of Justice and CINI.
- Open Relation Extraction on criminal sentences - Improving state-of-the-art for technical and legal texts. Use of text mining techniques, seq2seq models and Large Language Models
-
June 2022 - Aug 2022
Machine Learning Engeneer
Testudo S.R.L, Italy
- Worked with University MIND lab team developing and deploying a Machine Learning model to predict timestamps regarding working hours.
- I had the opportunity to work on a large amount of data and train a neural network to achieve excellent performance in a non-ideal case, improving productivity of the company's automatic systems.
-
March 2021 - May 2021
Intern
University of Milano - Bicocca, Italy
- University internship in the field of Hate Speech detection using NLP techniques on Tik Tok social network.
- Achieved excellent performance thanks to the use of lexicons and large language models with transformers-based architecture.
Academic Interests
-
Artificial Intelligence.
- Natural Languge Processing & Reinforcement Learning.
- Transformer-based architecture and generative language models.
Other Interests
- Hobbies: I'm Lazy Photographer and an active dreamer 🚀