Portfolio
On this page you can find some of my works that I enjoyed being a part of.
Towards the Automation of Responsible AI Evaluation Leveraging Large Language Models
My master thesis at AXA involved building LLM-driven evaluation pipeline that guides AI project teams through the process of evaluating their models, identifying ethical and legal risks, selecting appropriate fairness metrics and bias mitigation strategies, generating code implementations, and creating documentation that summarizes the interventions.
Throughout the thesis, different LLM types, prompt engineering methods and quantization methods were used.
DISCO - DIStributed COllaborative Learning (2022)
DISCO is developed by Machine Learning and Optimization Lab at EPFL to make these technologies accessible for broad groups of people with its open-source and open-access policy.
As a semester project student in Machine Learning and Optimization Lab at EPFL, I updated the front-end of DISCO to create an intuitive and informative experience for users regardless of their knowledge on machine learning concepts. I used Vue and TypeScript to create a visual identity.