AISensei

An AI-Powered Teacher Assistant that automates assignment grading for teachers, generates personalized feedback for students, and flags students falling behind, expanding access to quality education.

Comment

Inspiration

I built this because teachers carry a heavy load of admin and grading, especially in big classes or institutes with few resources. As the social and economic gap grows, teachers end up spending too much time on repetitive work like grading, instead of teaching. The arrival of large language models gave me a real chance to use technology to help communities and make quality education more available by reducing teachers' burden and making feedback faster, and more personal for students.

What it does

AISensei automates making assignments, tracking who turned them in, and grading with AI for teachers. It creates short, useful feedback for each student so they know what to fix next. It integrates with Google Classroom so teachers can keep their current workflow. The system also flags students who are falling behind in class so teachers can step in early. All of this aims to make learning fairer, more connected, and more hopeful.

How I built it

  • Backend: FastAPI, SQLAlchemy, PostgreSQL
  • Frontend: Next.js, Tailwind CSS
  • Async jobs: Celery, Redis
  • Auth & Integrations: Google OAuth, Google Classroom API, JWT
  • AI: Google Gemini API
  • DevOps: Docker, Docker Compose

Challenges I ran into

  • Tuning prompts so grading is consistent and fair across many question types.
  • Making async processing reliable, avoiding duplicate or missed grading jobs.
  • Respecting student privacy and handling Google Classroom token flows correctly.
  • Fixing CORS and other cross-origin problems during development.

Accomplishments that I'm proud of

  • A working LLM-driven grading loop that reports progress through WebSocket updates for near-real-time feedback.
  • A modular design that can be extended to other languages and subjects easily.
  • Built-in monitoring and retry logic that makes background grading powerful.

What I learned

  • Practical patterns for running background queues at scale and handling retries.
  • How to combine LLM outputs with clear rubric rules so grading stays trustworthy.
  • Best practices for OAuth with Google Classroom and how to store tokens securely.

What's next for AISensei

  • Multilingual grading so students can get feedback in regional languages and inclusion improves..
  • Adaptive learning paths that use feedback to create short, personalized study plans.
  • Versions for coaching institutes and corporate training for larger deployments.
  • An offline grading fallback so teachers in low-connectivity areas can still use the system.

Built With

Share this project:

Updates