An intelligent, interactive, Socratic AI Coding Tutor built as an 8-week internship capstone project.
This repository contains an advanced AI Coding Tutor Agent that guides students through Python programming errors using the Socratic method rather than giving direct answers.
The project is built on a Service-Oriented Architecture orchestrated by a LangGraph State Machine.
- LLM Layer: OpenAI / Groq (Prompt Engineering, JSON mode)
- Execution Layer: Local Subprocess Sandbox
- State Layer: ChromaDB (RAG/Memory) and SQLite Checkpoints
- UI Layer: Streamlit
- Evaluation Layer: Automated LLM-as-a-judge MLOps pipeline
- Week 1: Python Foundations
- Week 2: Prompt Engineering (Socratic Constraints)
- Week 3: Code Sandbox & Tool Use
- Week 4: Streamlit UI
- Week 5: ChromaDB Memory & Learner Profiles
- Week 6: LangGraph Orchestration & Human-in-the-Loop
- Week 7: MLOps Evaluation Framework
- Week 8: Capstone Architecture & Production Readiness
\�ash pip install -r requirements.txt streamlit run week4_mini_tutor/day4_streamlit_ui/app.py \
Developed as an AI Engineering Capstone.