Telepoisk
Search Engine for Telegram Chats
Overview
I've often run into situations where important, practical information can't be "just Googled" or obtained as a one-size-fits-all answer from ChatGPT—especially when it comes to healthcare, travel, restaurants, career decisions, business, or legal questions. In those moments, what matters most to me is real people's lived experience: their discussions, clarifications, nuance, and the freshness of the information.
To solve this, I built a search engine for public and private Telegram chats. The system is a distributed backend that scans hundreds of millions messages across a curated collection of chats, structures the data, and indexes the content for fast search.
Technologies
Backend: Django • PostgreSQL • Elasticsearch • Kafka
ML/AI: PyTorch • CUDA
Architecture: Microservices • Distributed Systems • Event-Driven
Languages: Python • C#
Detailed information coming soon