EduPilot AI

• 

  AI Smart Virtual Classroom: Real-time doubt solving and interactive learning powered by Gemini.

• 

  AI Continuity Teacher: Seamlessly taking over classes with instant lesson plans when teachers are absent.

Inspiration

Unexpected teacher absences disrupt the daily school schedule. We wanted to build a system that prevents this disruption by automating the substitution process and keeping the administration informed in real-time.

What it does

EduPilot AI functions as a continuity system for schools. When an administrator marks a teacher as absent, the system automatically:

• Analyzes the current syllabus and generates a structured lesson plan.
• Creates difficulty-based homework assignments.
• Initiates a Smart Virtual Classroom where a voice-synchronized AI delivers the lesson directly from uploaded textbook PDFs.
• Updates a centralized Principal's Dashboard with active AI classes and attendance intelligence, triggering automated alerts to parents.

How we built it

We divided the architecture into two main collaborative streams:

• Frontend & Backend (Owais Khan): Built the user interface using React and Tailwind CSS, and structured the backend logic using FastAPI and Node.js.
• AI Engine (Obaida Gul): Developed the core AI continuity logic and prompt engineering using Python, integrating Llama 3 via the Groq API.

Challenges we ran into

One of the main technical hurdles was parsing responses from the LLM. The model frequently attached markdown formatting to JSON outputs, which caused validation errors (json_validate_failed) in the backend. We resolved this by implementing a custom cleaning script to sanitize the raw content before parsing. We also had to ensure the voice engine synced accurately with the generated lesson text without significant latency.

Accomplishments that we're proud of

We successfully integrated the AI engine with the operational dashboard so that every automated action (like a generated lesson plan or a parent alert) reflects instantly on the administrative end.

What we learned

We learned how to manage a monorepo effectively, handle secure environment variables for deployments, and write strict parsing functions to ensure reliable data flow between a frontend, a Python backend, and an external LLM API.

What's next for EduPilot AI

Expanding the Attendance Intelligence module to include deeper predictive analytics for identifying students at risk of falling behind academically due to irregular attendance patterns.

Built With

• chromadb
• fastapi
• google-gemini
• groq
• mongodb
• node.js
• python
• react
• render
• sentence-transformers
• tailwind-css
• vercel
• vite