PinnyaPrep

Equal Educational Access for Every Student

Comment

Inspiration

In Myanmar, the Grade 12 national exam is a life-defining milestone. However, students face a massive "resource gap." While students in urban areas can afford expensive private tutors, those in rural areas struggle to find quality practice materials. Existing AI tools are often Western-centric and don't understand the specific nuances of the Myanmar curriculum or the Burmese language. We built PinnyaPrep to bridge this gap, acting as a tireless digital mentor for every student, regardless of their background.

What it does

PinnyaPrep is an AI-powered exam preparation platform specifically tailored for Myanmar’s national curriculum.

Smart Paper Generation: Instantly generates board-style practice questions based on specific subjects and chapters.

Bilingual Support: Understands and processes both English and Burmese to ensure no student is left behind by a language barrier.

Instant Feedback: Unlike a static PDF, PinnyaPrep explains why an answer is correct, mimicking the guidance of a personal tutor.

Curriculum Alignment: Specifically tuned to the formatting and style of Myanmar’s matriculation exams.

How we built it

We leveraged a modern tech stack to ensure the platform is lightweight and accessible:

LLM Integration: We used Large Language Models (LLMs) with custom system prompting to enforce strict adherence to the Myanmar Department of Education’s curriculum standards.

Retrieval-Augmented Generation (RAG): To ensure accuracy, we fed the AI a database of verified textbook content and historical exam patterns.

Frontend: Built with a mobile-first approach (using React/Next.js) because most students in Myanmar access the internet via smartphones.

Backend: A Python-based API to handle the logic of question randomized and evaluation.

Challenges we ran into

Formatting: Making sure math equations and Burmese script rendered correctly across all mobile devices was a significant UI hurdle.

Data Scarcity: Converting physical textbooks and old, blurry PDF past papers into machine-readable data required extensive pre-processing.

Accomplishments that we're proud of Accuracy: Successfully generating questions that experienced teachers confirmed "look and feel" like the real national exam.

Inclusivity: Creating a tool that works smoothly in Burmese, breaking down the English-only barrier common in EdTech.

UX/UI: Building an interface that is intuitive enough for a student to start practicing within two clicks.

What we learned

We learned that a general-purpose AI isn't enough for education; you need deep "domain context" (local curriculum) to be truly useful.

We discovered how to optimize LLM calls to keep costs low, which is essential for making the tool free or affordable for students.

What's next for PinnyaPrep

Offline Mode: Developing a way to download practice sets for students with unstable internet connections.

Progress Tracking: An analytics dashboard for students to see which chapters they need to focus on most based on their practice scores.

Community Features: A peer-to-peer discussion forum where students can help each other solve the AI-generated papers.

Built With

Share this project:

Updates