← All projects
AI · content platform2026 / Contora

From a news feed to a publishable story.

Source aggregation, deduplication, translation and publishing in one workflow.

My role

Architecture, web interface, source processing and AI integrations.

Scope

Contora

LinksOpen product
01 / The challenge

Bring scattered sources together and give editors a workflow for preparing and publishing stories.

02 / The solution

Contora

Source aggregation, deduplication, translation and publishing in one workflow.

  • Multi-source news aggregator
  • AI translation & rewrites
  • One-click Telegram autoposting
03 / The outcome

Editors manage the feed through the web dashboard and Telegram: reviewing stories, choosing processing steps and publishing to their channels.

Aggregation and clusteringTranslation and rewritingTelegram publishing
Technology
Implementation details
  • Implementing on-the-fly duplicate detection and news clustering algorithms using feedparser and rapidfuzz library token comparisons.
  • Designing strict prompt parameters for LLM APIs to ensure reliable structured output (JSON schema extraction) and high-quality multilingual rewriting.
  • Synchronizing python background worker tasks (APScheduler) with a Supabase PostgreSQL database and real-time frontend states in Next.js.
Next projectLoreCraft

Your next good product starts here

Have a project?
Let’s figure it out.

Bring an idea, a prototype or a process you have wanted to simplify.

Message on Telegram Build a brief

A short brief helps frame the conversation.
No commitment to start a project.