I've always found writing research papers to be painfully chaotic. I'd have hundreds of tabs open, download piles of PDFs, and still spend hours digging through each one just to find that one relevant paragraph. It was frustrating, inefficient, and honestly a motivation killer.
So I built Synthara to make that process easier.
Synthara is a research assistant workspace where I can upload any file related to my project—papers, slides, notes and instantly turns them into a searchable conversational knowledge base. Synthara uses semantic search powered by Retrieval-Augmented Generation. Synthara understands my queries and pulls up exactly what I need, without me having to read every single line. It's made extracting information for my research so much faster, more focused, and actually enjoyable.
The app is built with Next.js and styled using Tailwind CSS for a clean, responsive interface. I used tRPC to create a fully type-safe API layer, and Prisma with PostgreSQL (NeonDB) to manage structured data. For semantic search, I integrated Pinecone to store and query vector embeddings, enabling fast and accurate retrieval. User authentication is handled with Kinde, making it secure and ready for multi-user access.
