Inspiration
At our college, Gondwana University, finding a single question paper feels harder than the exam itself. The official portal is a time-sink of messy folders and slow loading. Currently, a student has to download a massive 80MB+ bulk ZIP file containing thousands of unrelated papers just to find one 2MB PDF. We were inspired to kill this manual grind and replace it with a high-speed, autonomous "Resource Hunter."
What it does
PYQ Portal is a student-led initiative that transforms a broken system into a lethal productivity tool. It bypasses the university's chaotic UI to deliver the exact paper you need in under 10 seconds. With a streamlined 5-step selection process (Session ➔ Branch ➔ Semester ➔ Subject), it provides direct PDF access without the burden of bulk downloads. It is fully compatible with all engineering branches and the latest NEP/CBCS curriculum patterns.
How we built it
We forged a Stateless Proxy Architecture designed for speed and zero maintenance: Backend: Python and Flask act as a high-speed middleware that hits the university’s AJAX endpoints directly. Automation: We implemented pattern-matching algorithms to identify ZIP naming conventions and predict future sessions (W2025, S2026). Database: A robust SQLite system manages metadata, while a processing pipeline extracts and parses filenames into structured categories. Frontend: An industrial 'LavaPunk' dark theme built with Tailwind CSS, optimized for mobile-first campus usage.
Challenges we ran into
The university’s legacy infrastructure was the biggest beast. The site uses dynamic content that often breaks standard scrapers. We had to reverse-engineer the network logs to find the hidden customers_mysql.php endpoint. Additionally, making the system "Server-Optimized" to process 500MB ZIP files without crashing free-tier hosting required intensive memory management and efficient extraction logic.
Accomplishments that we're proud of
We successfully turned a 10-minute struggle into a 10-second click. Our system achieved a 1.3-second detection speed, making it 60x faster than the manual method. We are proud to have built a tool that is already saving thousands of student-hours for our campus. Winning the Kiroween hackathon previously gave us the momentum to scale this into the autonomous beast it is today.
What we learned
We learned that technical complexity doesn't matter if it doesn't solve a human problem. This project taught us about Institutional Impact—how a small piece of automation can remove a massive bottleneck for an entire community. We also mastered the art of building "Stateless" systems that can live and breathe in sync with a third-party legacy portal without constant manual updates.
What's next for Pyq portal
The hunt doesn't stop here. We are moving toward a Conversational AI Agent that allows students to query papers using natural language. Imagine asking, "Find me the most repeated question on Normalization from the last 3 years," and getting an instant answer. We also plan to scale this architecture to support other universities facing similar "Bulk ZIP" nightmares. "We didn't just build a portal. We engineered a shortcut through the chaos."


Log in or sign up for Devpost to join the conversation.