A platform that reads the news and scores what matters
An upskilling and networking app that scrapes news at scale, then ranks each article by the entities and relations inside it using a knowledge graph — so people see what's actually worth their time.
- Generated personalised learning content with LLMs — summaries, text-to-speech, podcasts, and quizzes.
- Clustered users by goal: SentenceTransformers embeddings grouped with Ward's hierarchical clustering, plus a custom threshold that finds the cleanest communities — matching people for debates and discussions.
- Ran heavy jobs in parallel with a hybrid thread/process pool, and exposed everything through FastAPI: users, content, messaging, recordings, automatic recaps.
- Knowledge Graphs
- RAG
- SentenceTransformers
- Hierarchical Clustering
- LLMs
- Web Scraping
- FastAPI
- MongoDB
- MinIO
- Python