Projects – Lucas Sylvester

By - Lucas Sylvester
Posted on May 4, 2025September 3, 2025
Posted in Projects

Winning the 2025 Demant – Audio Explorers Engineering Case Competition

To challenge ourselves and as fun team project me and some friend decided to compete in the yearly Demant Audio Explorer challenge. The challenge is an engineering case competition, with cases aligning with different domains and areas of study. The winners of each case win a free trip to New York, where they partipate in

By - Lucas Sylvester
Posted on December 14, 2024December 26, 2024
Posted in Projects

Efficient training of cross domain LLMs with LoRA and model merging

By - Lucas Sylvester
Posted on October 5, 2024October 5, 2024
Posted in Projects

Volunteer project: Providing accessible and free servers for students.

At the start of the semester, I launched an exciting initiative aimed at giving students free access to servers for hands-on learning and exploration. This project had been on my mind for a while, and I finally had the chance to bring it to life. DTU is well-equipped with computing resources, especially through the high-performance

By - Lucas Sylvester
Posted on December 22, 2023May 4, 2025
Posted in Projects

Exploring semantic differences in book reviews on female-led and male-led Young Adult novels

As part of my work with COMP550 Natural Language Processing, we got to undertake a major project exploring reader perception of characters in young adult Novels using the word2vec and the SemAxis framework. The project sought to investigate both typical gender stereotypes as well as perception of similar traits in men and women by looking

By - Lucas Sylvester
Posted on October 1, 2023November 3, 2023
Posted in Projects, Travel

MAIS Hacks 2023 – Winner of Hack for Impact

This past weekend I had the amazing opportunity to participate in my first hackathon: MAIS Hacks organized by the McGill Artificial Intelligence Society My teammates Roman Sejnoha, Murad Ismayilov, Jasneet Sawhney and I had a great time developing our TimeTrek browser extension over the course of 24 hours. Using natural language processing and machine learning, we put together and tested a