Mohamed Amine Hajji / Projects

Research Portal

A full-stack research-publishing platform with role-based access and a per-paper AI chat, grounded in the paper’s own text.

Research Portal is a full-stack platform for publishing, discovering, and discussing research papers, with role-based access for students, researchers, and general users. Each paper gets its own AI chat — grounded in the paper’s extracted text and powered by Llama 3.3 70B via Hugging Face — so readers can ask questions and get contextual answers instead of skimming the PDF cold. The backend is a layered Node/Express API (Routes → Controllers → Interactors → Repositories → Models) wired together with Inversify dependency injection, sitting on PostgreSQL via Sequelize.

Highlights

• Per-paper AI chat (Llama 3.3 70B via Hugging Face) grounded in the extracted PDF text, for contextual Q&A instead of blind skimming.
• Layered backend architecture — Routes → Controllers → Interactors → Repositories → Models — wired with Inversify dependency injection.
• Automatic PDF thumbnail generation, role-based access control, and a Chart.js analytics dashboard, backed by Swagger-documented REST APIs.

Stack: Angular, TypeScript, Node.js, Express, PostgreSQL, Sequelize, Hugging Face, Docker