Socrates AI – Conversational Agent with Prompt Evaluation & Session Memory
Built an AI-powered learning assistant in Python that uses natural language processing to interact with users, optimize task schedules (Pomodoro method), and summarize sessions. Developed custom logic for prompt evaluation, intent tagging, and memory management using tiktoken to minimize context length while preserving semantic continuity. Worked with fine-tuned prompt–response pairs to simulate and label user input for clarity, emotion, and task priority classification.