👋🏽 Andrew Koulogeorge
📚 I am a first year Masters in Computer Science student at Carnegie Mellon University. I am focusing my coursework at Carnegie Mellon on Machine Learning Theory & Machine Learning Systems.
💼 This summer, I worked as an Applied Scientist Intern at harpin AI where I applied Neural Networks and Text Embedding Models to improve the companies core identity profile similarity model. I was very fortunate to be mentored by Max De Jong and Wenzhong Zhao.
🎓 This past June, I graduated from Dartmouth College Summa Cum Laude in Mathematics and Computer Science.
🔬 While at Dartmouth, I worked as a Deep Learning Researcher at the Minds, Machines and Society Group where my work focused on improving interpretability and preventing bias in Large Language Models. I was lucky to have two excellent advisors in Soroush Vosoughi and Yoaqing Yang!
📝 In my current research project, I identified critical flaws in existing Prototypical Large Language Model architectures that were leading to unfaithful explanations. To address these issues, I developed a novel framework called Faithful Alignment, which corrects these architectural shortcomings across various Prototypical Language Models. My findings demonstrate that Faithful Alignment successfully restores the faithfulness of model explanations without compromising accuracy. This work is currently under review and a preprint will be released soon!
🔍 I am actively looking for internship opportunities in Machine Learning & Software Development for Summer 2025!