After months of deep research, engineering, testing, and refinement, our team is beyond proud to officially present AutoGradePro — an AI-powered, production-grade platform that automates and enhances grading for educators. 🔍 What is AutoGradePro? AutoGradePro is an intelligent grading system built to: ✅ Upload and process 200+ student answer scripts in bulk ✅ Automatically extract questions and marking schemes from .pdf, .docx, and .xlsx files and store them in PostgreSQL as structured JSON ✅ Fetch structured marking data to the client for seamless interaction ✅ Use LLMs (Ollama – Qwen2.5:7B) for semantic answer evaluation — not just keyword matching ✅ Grade diverse answer formats: one-word, short phrase, list-based, and numerical-range responses ✅ Deliver fair, AI-assisted scoring for open-ended and semantically varied student answers ✅ Generate instructor-level analytics and performance reports ✅ Support multiple file types: .PDF, .DOCX, .TXT, .XLSX ✅ Fully dockerized, deployed, and performance-tested for concurrent grading requests ⚙️ Tech Stack 🛠 Django | Next.js | Tailwind CSS | Ollama (LLMs) | Docker | AWS (m6i.large) | PostgreSQL 💬 Client Feedback That Meant Everything We were honored to work under the guidance of Dr. Nadeesha Sandamali, Senior Lecturer at the University of Ruhuna. Her mentorship pushed our project far beyond what we thought possible. Here’s an excerpt from her feedback: I personally tested the system using my own question set and over 40 student answer scripts, and found the overall performance of the system to be reliable and effective. The integration of AI for automated evaluation was particularly impressive. The team delivered the project well beyond my expectations with great responsiveness to feedback. I look forward to using this system to streamline the marking of continuous assessments in upcoming modules. We are truly grateful for her trust, wisdom, and continuous support throughout this journey.