Knowledge RAG Bot
Production-ready Telegram bot and FastAPI backend for personal knowledge retrieval with citation-aware, grounded answers.
What it is
Retrieval-augmented Telegram bot where you upload PDFs, audio, or text and ask grounded questions — answers cite exact source passages.
My role
Builder across ingestion pipeline, retrieval strategy, FastAPI backend, Telegram bot interface, and prompt grounding.
Timeline
~2 to 3 weeks
Stack
Python, FastAPI, Qdrant, MongoDB, Telegram Bot API, OCR, LLM
Technical Workflow Simulator
Explore the core engineering mechanism through a focused product walkthrough.
9:41
Walkthrough ready
Scroll the device into view to start the demo.
Knowledge RAG assistant
- 1
Bot ready
The Telegram bot waits for a /query_doc command.
- 2
Question sent
- 3
Vector search
- 4
Grounded prompt
- 5
Cited answer
Bot ready
The Telegram bot waits for a /query_doc command.
The Problem
Generic chatbots are weak when answers need to come from specific documentation or internal knowledge. The goal was to build a system that can cite grounded context and reduce hallucination risk — accessible directly from Telegram.