Shaping the Future of Education: Advancing Personalized AI Learning Companion™
Every day, 1.5 billion students encounter challenges with their homework, and Brainly is here to assist them. One of the platform’s primary goals is to provide widespread access to personalized learning. With the rapid development of AI, particularly large language models, new tools emerge daily, bringing us closer to this objective. This talk presents a brief overview of research focused on learning personalization through machine learning. The first part will cover Curriculum Content Retrieval, tracing its evolution from early concepts to full-scale production deployment supporting all of Brainly’s traffic. In the second part, we will demonstrate how we leverage curriculum content to power our Retrieval-Augmented Generation (RAG) system, surpassing traditional RAG approaches. Techniques such as prompt engineering and model fine-tuning will further advance our personalized AI Learning Companion™.
Bio
Andrzej has graduated from Aalto University and AGH in machine learning. Throughout his career, he was mostly involved in computer vision & NLP related problems. At Brainly, he works in the personalization domain.