FAMA is a full-scale pipeline designed to analyze public opinions at scale. It takes a user's question, scours Reddit for relevant discussions, extracts content, semantically matches it, and generates structured, actionable reports.
I needed a way to quickly understand public sentiment on niche topics without manually reading hundreds of posts. Most existing tools were too shallow, so I built a system with research depth in mind.
Defined the complete flow: extraction, cleaning, semantic scoring, clustering, and summarization.
Integrated semantic search models to identify high-relevance content, filtering out noise.
Created a layer that generates executive summaries, identifies key themes, and highlights contradictions.
Built a "Chat Mode" allowing users to directly interrogate the extracted data for deeper insights.
This project demonstrates my ability to design research tools end-to-end. I didn't just build an interface; I architected a full system that handles data ingestion, complex analysis, and usable output.