Deployed & in use
Echo Shield
AI Social Narrative Analysis & Counter-Intervention Platform
Echo Shield watches public conversation across major social platforms and makes sense of it at scale. A multi-stage NLP pipeline reads each post — detecting language, entities, sentiment, and stance — then groups related content into the narratives quietly forming underneath the noise.
Once a narrative surfaces, a small crew of AI agents studies it and drafts a considered counter-narrative response. The whole system runs its own machine-learning models locally rather than leaning on hosted APIs.
//what makes it work
Real ML, end to end
Language detection, entity recognition, sentiment, stance, embeddings, and clustering all run as local model inference — including support for Urdu and Roman Urdu, not just English.
Sees past text
Media enrichment pulls text out of images and speech out of video before analysis, so a narrative carried in a screenshot or a clip isn't invisible to the pipeline.
Agents that respond, not just report
A multi-agent system moves from analysis to action — studying a narrative and drafting a strategic counter-response rather than stopping at a dashboard.
built with
- Python
- FastAPI
- Local LLM
- React / TypeScript
- NLP models