Galápagos is a tool that automates the generation of N-Version program variants using LLMs and validates their equivalence, demonstrating effectiveness in creating diverse and reliable variants, even across different programming languages.
Collaborating alongside a group of colleagues with SEB, a leading Nordic bank, on a joint project for my Project Course in Data Science at KTH. My role involves proposing methods to enhance accuracy by integrating multiple large language models (LLMs) in a consensus-based pipeline. Our efforts also include dataset acquisition, data cleaning, and preprocessing, as well as managing the technical setup and integration of the language models within the project.