Solving AI Safety with a Quality Assurance platform for *all* AI models:- Co-developed an end-to-end RAG pipeline for policy compliance to lead a Proof of Concept.- Built a pipeline to summarize code and PDF files, images and tables included, to improve retrieval by 20%.- Fine-tuned a LLM to improve a RAG’s F1-score by 15% with as little data as possible and wrote an article about it.- Researched fairness metrics and implemented mitigation techniques for the Giskard open-source library.- Wrote a technical article for Hugging Face to demonstrate the full potential of the Giskard scan.- Weekly presentations of the latest academic research papers to stay up to date with the current state of AI.

Use of Maya to create 3D models that were later used in movies.