Data and AI
Four Python systems: data collection from a hundred sources, content intake with moderation, a publishing platform with an SEO layer and a market data lab built with an agentic pipeline.
The problem
Data arrives from websites, open APIs, Telegram and exchanges — in different formats, with duplicates and outages. One-off scripts do not hold up here: sources change, parsers break, and errors quietly corrupt the data.
What is needed are services that collect, normalise and check data continuously and, where the result reaches people, pass it through moderation.
Technology
The solution
Collection from 100+ sources — websites, open APIs and Telegram — with normalisation, deduplication and failure monitoring, alongside FastAPI services for catalogue and data enrichment.
25 site-specific parsers on Celery, RabbitMQ and Playwright, with source-URL deduplication. Material flows through a REST contract into a moderation queue and is published after review.
FastAPI, async SQLAlchemy and PostgreSQL; a multi-theme Jinja2 front end and an admin CMS with an editorial moderation queue.
Four exchange APIs stream into TimescaleDB, with a rules engine of 40+ risk parameters on top. The system was built end to end with an agentic pipeline.
Outcome
The services run continuously: collection does not stop when one source changes, failures show up in monitoring, and material that reaches people goes through moderation.
Contacts
Tell us in two paragraphs what's not working. We will tell you whether we take it on and where we suggest starting. Meetings online, or in person in Uzhhorod or Kyiv.
