Soutify: Social Media Listening Platform
Near-real-time brand monitoring across Twitter and global news, with Arabic and English AI sentiment analysis and flexible Boolean search.
The problem
Clients needed to know what was being said about them across Twitter and global news as it happened, in Arabic as well as English, a pair most listening tools handle poorly or not at all.
Constraints
- Continuous multi-source ingestion without making the search experience unresponsive
- Arabic sentiment is not translated English sentiment: dialect and negation defeat naive models
- Users needed arbitrary Boolean queries over accounts, hashtags, keywords and time windows
Key decisions
Django for ingest and query chosen over a lighter async framework
the hard part was the query model, not the I/O. The ORM, migrations and admin carried a complex relational schema that would otherwise have been hand-written and hand-maintained.
Soutify helps clients track brand mentions, analyze public sentiment, and understand trending topics across multiple platforms simultaneously.
I designed and implemented the backend infrastructure using Django. The core challenge was continuously ingesting Twitter and global news feeds while keeping search and analysis responsive.
I built the monitoring system that lets users create custom search parameters: tracking specific users, hashtags, keywords, and complex Boolean queries across time periods. Working with the AI team, I integrated sentiment analysis for Arabic and English content with nuanced sentiment scores and topic classification.
