Apollo Search - AI-Powered Academic Research Assistant

Main screen
Advance Search
Saved papers
Papers for Citation

Project Story

From Frustration to Innovation: The Birth of Apollo Search

The Spark

It all began during my master research in human abnormal pose detection. After spending countless hours navigating through disjointed academic search engines, downloading papers from various sources, and manually synthesizing information, I realized there had to be a better way.

The turning point came when I discovered the power of AI-based search and synthesis. What if researchers could not only find relevant papers but also get instant summaries and connections between research findings?

The Journey

Building Apollo Search wasn't straightforward. The initial challenge was finding an AI model capable of understanding academic content with sufficient depth. After experimenting with several options, Perplexity's API emerged as the clear winner due to its ability to search and synthesize scholarly content with remarkable accuracy.

The technical implementation journey involved:

Setting up the Next.js framework - Choosing a modern, performant foundation
Designing the search interface - Creating intuitive advanced search options for academic parameters - with a SciHub fan
Integrating the Perplexity API - Moving from the SDK to direct API calls for better control
Building the results display - Crafting a UI that presents academic information clearly with proper citations
Enhancing with filters and sorting - Adding tools researchers need to refine their searches

Challenges Overcome

Several obstacles appeared along the way:

API Response Formatting: Perplexity's responses occasionally came in inconsistent formats, requiring robust parsing logic
Citation Extraction: Ensuring proper academic citations from AI-generated content was particularly challenging
Performance Optimization: Balancing comprehensive search results with response time
Academic Accuracy: Fine-tuning prompts to ensure scholarly precision in the results

The Result

Today, Apollo Search stands as a bridge between traditional academic research methods and AI-powered knowledge discovery. It transforms hours of manual searching into seconds of intelligent exploration, helping researchers focus on what truly matters - understanding and building upon existing knowledge.

The project continues to evolve, with plans for multi-language support, expanded citation styles, and deeper integration with reference management tools.