Telescope - A research assistant

Demo

A full-stack web application for searching and interacting with space-related research papers. Built for NASA Space Apps Challenge, this project provides full-text search capabilities over a corpus of scientific PDFs with an AI-powered chat interface.

Features

• Full-Text Search: SQLite FTS5-based search across indexed research papers with metadata (titles, authors, publications)
• AI Chat Interface: Interactive AI assistant that can search through papers and answer questions using retrieved context
• Modern Web UI: Built with SolidJS and TailwindCSS for a responsive, fast user experience

Tech Stack

Backend (Rust)

• Axum for HTTP API server
• SQLite with FTS5 for full-text search
• WebSocket server for AI streaming
  • This is not used
  • I wasted 2 hour writing this before I figured out I didn't need it
• PDF text extraction and indexing
• The backend/scripts directory builds the search index. Run npm download then npm index to build the search index
• Deployable with docker see Justfile for the docker commands

Frontend (TypeScript/SolidJS)

• SolidJS for reactive UI
• OpenAI integration for chat
• TailwindCSS for styling
• Markdown rendering for formatted responses

Database

The project uses SQLite FTS5 a pre-indexed database of 680 space-related research papers with full metadata (accessible via /backend/search_index/).