🤖 Phase 2: AI Tutor - Contextual Q&A System

## Phase 2: RAG Query Engine

Build the "Ask the Video" feature with context-aware responses.

### Tasks
- [ ] Create RAG Query Endpoint: POST /ask-tutor
- [ ] Implement semantic search over vector database
- [ ] Build context assembly from retrieved chunks
- [ ] Design prompt template for contextual Q&A
- [ ] Add citation system (timestamp links to video)
- [ ] Implement conversation memory for follow-up questions

### Acceptance Criteria
- Users can ask questions about current video content
- System retrieves relevant transcript segments
- Responses include timestamps for source verification
- Supports follow-up questions with conversation context
- Graceful handling when content not found in transcripts

### Tech Stack
- Vector similarity search (cosine similarity)
- Gemini/GPT-4 for answer generation
- RAG prompt engineering
- Conversation history management