AcademiaAI

Turn complex academic notes and research papers into structured study guides, vocabulary breakdowns, and practice quizzes powered by Gemini Flash and Flask.

Comment

Inspiration

As students and researchers dealing with heavy academic workloads, we frequently find ourselves drowning in dense research papers, textbook chapters, and lengthy lecture notes. Synthesizing complex subjects—like biochemistry, computer science, or medicine—by hand is incredibly time-consuming. Traditional text summarizers often produce unstructured blocks of text that are difficult to revise from. We wanted to build a tool that doesn't just shorten text, but transforms it into a structured, active-recall study kit instantly.

What it does

AcademiaAI is an AI-powered academic assistant that converts complex, unstructured text into an organized, high-yield study kit in seconds. Key features include:

  • 📌 High-Yield Concept Summaries: Condenses verbose literature into structured, actionable bullet points.
  • 💡 Essential Vocabulary & Definitions: Automatically extracts domain-specific terminology for quick memorization.
  • 📝 Self-Assessment Quizzes: Generates Multiple-Choice Questions (MCQs) with an answer key for active recall practice.
  • ⚡ Real-Time Formatting: Renders clean, client-side Markdown using Marked.js for effortless reading.

How we built it

  • Backend: Developed with Python and Flask to handle lightweight request routing and prompt orchestration.
  • AI Engine: Powered by Google's Gemini Flash model (google-genai SDK) to deliver structured academic breakdowns.
  • Frontend: Built using modern HTML5, CSS3, and JavaScript, using Marked.js for dynamic Markdown rendering.
  • Database & Analytics: Integrated with ClickHouse Cloud (clickhouse-connect) to asynchronously log topic telemetry for usage tracking.

Challenges we ran into

  • Structured AI Outputs: Engineering prompts that consistently enforce strict Markdown sections (Summaries, Vocabulary, and Quizzes) without breaking syntax or hallucinating details.
  • Non-Blocking Telemetry: Ensuring ClickHouse database calls run asynchronously so network latency never delays delivering instant study kits to the user.
  • Corrupted Virtual Environments: Managing local deployment dependencies and terminal environment issues during rapid iteration.

Accomplishments that we're proud of

  • Successfully creating an end-to-end web application that processes raw academic input and outputs a formatted study guide in under 3 seconds.
  • Achieving clean, real-time client-side rendering for complex Markdown elements like headers, lists, and quiz formats.
  • Seamlessly linking Google Gemini Flash with ClickHouse Cloud telemetry in a single Flask backend.

What we learned

  • Effective prompt-engineering techniques for enforcing structured Markdown output from LLMs.
  • Asynchronous error handling and connection management with cloud databases in lightweight web servers.
  • Efficient frontend parsing and dynamic DOM updates using Vanilla JavaScript and Marked.js.

What's next for AcademiaAI

  • Multimodal Uploads: Support for direct PDF research paper parsing and textbook page image scans.
  • Flashcard Export: One-click exports of generated vocabulary and quizzes to Anki and Quizlet.
  • Analytics Dashboard: Personal progress tracking based on study history stored in ClickHouse Cloud.

Built With

Share this project:

Updates