setos.ai Startup Prototype & Pitch

A startup prototype that turns a plain-language question into a reading path through more than a million research papers.

• Led a 3-person team building personalized learning pathways across 1,000,000+ research papers from natural language queries.
• Designed a content-based recommender with SciBERT embeddings, pgvector similarity search, LLM-based query expansion, and the kneedle algorithm to auto-tune pathway length.