Study-Pilot

Inspiration

Students often struggle to organize learning materials, identify knowledge gaps, and decide what to study next. I created StudyPilot to transform passive AI assistance into an active, personalized learning workflow.

What it does

StudyPilot converts PDFs, presentations, notes, audio, and videos into a personalized study experience. It provides source-cited tutoring, adaptive quizzes, mock exams, flashcards, spaced repetition, and progress tracking. It identifies weak topics and recommends what to revise next.

How we built it

I built StudyPilot using Python, FastAPI, JavaScript, and OpenRouter for AI capabilities. Document processing, OCR, speech transcription, and source retrieval connect uploaded materials to tutoring and assessment. Learner tracking records mistakes and topic mastery to guide revision. Configurable model fallbacks and source-based practice keep essential features available when AI services fail.

Challenges we ran into

My biggest challenges were grounding AI responses in actual course materials, handling different file formats, and managing API timeouts and rate limits. I addressed these through source citations, retrieval-based answers, request budgets, fallback models, and locally generated concept questions.

Accomplishments that we're proud of

I integrated material ingestion, source-linked tutoring, adaptive assessments, a daily study cockpit, a mistake notebook, and revision tools into one application. StudyPilot can connect learning goals with actionable study tasks instead of leaving students to plan everything themselves.

What we learned

I learned how to integrate language models, information retrieval, databases, document processing, and learner-performance tracking into a practical AI system. I also learned that reliable AI requires verification, graceful failure handling, and grounding answers in trustworthy sources.

What's next for Study-Pilot

I aim to strengthen autonomous study planning, improve mastery predictions, personalize revision using long-term learning patterns, and introduce richer integrations that help students move from setting a learning goal to completing and evaluating a full study session.

Built With

• agentic-ai
• css
• fast-api
• faster-whisper
• html
• information-retreival
• javascript
• numpy
• ocr
• openrouter
• pymupdf
• python
• python-pptx
• scikit-learn
• sqlite