TrendNest

TrendNest is a trend discovery tool that gathers weak signals from multiple sources and organizes them into topic views for product, content, and campaign decisions. It keeps ingestion quality, clustering, and presentation aligned so the signal stays readable without pretending it is more certain than it is.

Trend research often ends up split across spreadsheets, screenshots, and intuition, which makes it hard to compare sources or explain why something matters.

I built a service that ingests mixed signals, groups related items into topic clusters, and presents them through a reviewable interface with source-level traceability.

Python / FastAPI / PostgreSQL / Embeddings / Next.js

• Combined structured and unstructured source ingestion in one pipeline.
• Used embedding-based grouping to reduce duplicate topic discovery.
• Stored source references for each cluster to support manual review.
• Scored velocity, recurrence, and cross-source agreement.