The project
WebTutorCenter helps users search for tutors by subject and area, publish class requirements and manage tutor applications. The platform supports learners, tutors and administrators through separate workflows and permissions.
Core experience
- Tutor discovery, filtering and detailed public profiles.
- Class posting, fee quotation, applications and direct tutor invitations.
- Candidate selection, admin approval and class status transitions.
- Realtime user-to-admin chat, notifications, reviews and personal vouchers.
- Responsive web experience with a companion mobile repository.
My contribution
I led a three-person team and focused on backend architecture while contributing across the full stack. I designed REST APIs and MongoDB models for users, tutors, classes, applications, reviews, promotions and notifications.
I implemented the matching workflow from application or invitation through selection and approval, along with validation, authorization, sensitive-data masking and class lifecycle handling.
AI chatbot service
I developed a standalone Python and FastAPI chatbot service using async HTTP requests, JWT forwarding, internal-service authentication, timeouts and safe error handling.
The response pipeline combines cached personal and live data, keyword and FAQ resolvers, Vietnamese text normalization and ranked RAG-lite context before using Groq Llama 3.3 as a fallback with confidence scoring.
Engineering details
The platform also uses Mongoose, Joi, JWT refresh tokens, Google OAuth, Socket.IO, Redux Toolkit, Tailwind CSS, pytest, Docker and scheduled background jobs.