I want to build an AI-powered tutoring experience aimed at adults (18+) who already grasp the basics and now need intermediate-level guidance. The exact subject area can remain flexible for the first release—what matters is that the engine can adapt its explanations, examples, and exercises to a learner who is beyond beginner yet not quite advanced. Core objectives • Conversational tutoring: natural-language Q&A where the model clarifies concepts, offers examples, and quizzes the learner. • Adaptive pathway: difficulty and pacing adjust automatically as the student demonstrates mastery or confusion. • Progress insight: simple dashboards that show strengths, gaps, and recommended next steps. Technical expectations The solution may draw on Python, TensorFlow/PyTorch, or comparable frameworks, and can leverage an LLM API for dialogue. A lightweight UI—web or mobile—is enough for this phase as long as it cleanly surfaces the chat and progress features. Deliverables 1. A working prototype ready to deploy for limited beta use. 2. Source code, commented and organized for future extension to other subjects. 3. Brief implementation guide explaining model choices, data flow, and how to retrain or fine-tune for new material. Acceptance criteria • Handles intermediate questions with at least 90 % correct or helpful response rate in a 50-item test set. • Adjusts question difficulty within two turns of detecting over- or under-performance. • Stores and displays session history and progress without data loss. I will be available for quick feedback cycles, and I’m happy to supply domain content once the core logic is in place.