
Telegram University Assistant Chatbot
Campus Information Assistant Inside Telegram
“Reliable bots need layered fallback, not a single answer path.”
System Overview
This project is a Telegram long-polling FAQ assistant built in Python with python-telegram-bot. Incoming messages are routed through an orchestration layer that attempts multiple answer strategies before returning a safe fallback response.
Layered Retrieval Pipeline
The core flow prioritizes a robust rule-based FAQ system for high-confidence answers, then escalates to an intelligent search agent and crawler-assisted context building. If confidence remains low, it synthesizes a response through local model-serving endpoints with OpenAI-compatible and Ollama-style fallback calls.
Implementation Notes
The codebase includes strong architectural intent for retrieval-augmented answering, but there are currently interface mismatches between some modules and expected data shapes. Even with this, the layered design demonstrates a practical approach to balancing reliability, dynamic retrieval, and model generation in chat assistants.
Project Gallery
2 items

Explore More Projects
Swipe or click arrows to see other work
















