Abstract

In today's rapidly evolving professional landscape, individuals must continuously update their skills to remain competitive. However, traditional educational systems and static e-learning platforms often fail to provide personalized learning paths tailored to each professional's goals, background, and pace. To address this challenge, we present SaarthiAI, an AI-driven adaptive learning system designed to generate customized professional learning plans and deliver targeted, interactive instruction. SaarthiAI integrates a Roadmap Generator leveraging retrieval-augmented generation (RAG) and dense vector retrieval via FAISS to construct personalized learning roadmaps from a knowledge base of industry-relevant content. It incorporates adaptive assessments powered by large language models to evaluate proficiency and dynamically adjust content difficulty. An AI Tutor chatbot module provides real-time contextual assistance and guidance. The system is implemented using Python, utilizing the Hugging Face Transformers library, MongoDB for data storage, and a RESTful API for seamless integration. Our contributions include the novel integration of RAG for roadmap generation, dynamic assessment mechanisms, and an interactive AI Tutor, collectively advancing personalized professional education.

References

• Zhu, X., Zhou, Y., & Xu, Z. (2022). Dynamic Curriculum Generation Using Neural Retrieval Models. IEEE Transactions on Learning Technologies, 15(3), 145–157.
• D Guu, K., Lee, K., Tung, Z., Pasupat, P., & Chang, M. (2020). Retrieval Augmented Language Model Training for Open Domain Question Answering. In Proceedings of ICLR.
• Karpukhin, V., Oguz, B., Min, S., Wu, L., Edunov, S., Chen, D., & Yih, W. (2020). Dense Passage Retrieval for Open-Domain Question Answering. In Proceedings of EMNLP.
• Izacard, G., & Grave, E. (2021). Leveraging Pre-trained Checkpoints for Sequence Generation Tasks. In Proceedings of NAACL.
• Embretson, S. E., & Reise, S. P. (2000). Item Response Theory for Psychologists. Psychology Press.
• Corbett, A., & Anderson, J. (1995). Knowledge Tracing: Modeling the Acquisition of Procedural Knowledge. User Modeling and User-Adapted Interaction, 4(4), 253–278..
• Brusilovsky, P. (2001). Adaptive Hypermedia. User Modeling and User-Adapted Interaction, 11(1-2), 87–110.
• Lee, S., Park, H., & Kim, Y. (2020). Engagement Metrics in E-Learning Platforms. Computers & Education, 143, 10368.
• Dziuban, C., Moskal, P., & Hartman, J. (2018). Bridging the Gaps: Blended Learning Approaches in Higher Education. Online Learning Consortium Journal, 12(4), 67–79.
• Johnson, M., & Brown, L. (2021). Data-Driven Adaptive Learning: Techniques and Frameworks. Journal of Learning Analytics, 8(1), 23–38
• FastAPI Contributors. (2021). FastAPI: The high-performance web framework for building APIs with Python 3.6+ based on standard Python type hints. Retrieved from https://fastapi.tiangolo.com/
• MongoDB, Inc. (2024). MongoDB: The Developer Data Platform. Retrieved from https://www.mongodb.com/
• Zenke, F., & Vogels, T. P. (2021). The remarkable robustness of surrogate gradient learning for instilling complex function in spiking neural networks. Neural Computation, 33(4), 899–925.
• Wolf, T., Debut, L., Sanh, V., Chaumond, J., et al. (2020). Transformers: State-of-the-art natural language processing. In Proceedings of EMNLP: System Demonstrations.
• Johnson, J., Douze, M., & Jégou, H. (2017). Billion-scale similarity search with GPUs. IEEE Transactions on Big Data, 7(3), 535–547.
• Lewis, M., Oguz, B., Rinott, R., Petroni, F., Khashabi, D., & Riedel, S. (2020). Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks. In Proceedings of EMNLP.