Guided problem solving
Instant evaluation and feedback
Goal based performance tracking
Question mining and generation
Answer evaluation engine

Aryabhata — Your Personal AI Prep Coach

Inspiration

We are living in a time where high-quality educational content is widely available online for almost every subject and competitive exam. From NCERT resources to JEE preparation materials, students have more access to learning content than ever before.

However, content availability has not solved the core problem: students still lack personalized guidance.

Many students struggle with:

Not knowing which topics to prioritize
Identifying weak areas
Tracking progress toward exam goals
Getting help while solving problems
Receiving realistic evaluation of mock tests

Personalized mentorship solves these problems, but access to good mentors and coaching is limited and often expensive.

We built Aryabhata to make personalized academic guidance scalable using AI.

What it does

Aryabhata — Your Personal Prep Coach is an AI-powered learning assistant that helps students prepare for exams through guided practice, automated evaluation, and progress tracking.

The system currently supports CBSE Grades 8–12 and JEE preparation.

Aryabhata provides three core capabilities:

Guided Problem Solving

Provides step-by-step hints while students solve problems
Encourages reasoning instead of revealing full solutions
Acts like a mentor sitting beside the student

AI Mock Test Evaluation

Generates realistic mock tests based on syllabus and topic weightage
Allows students to upload answer sheet images
Evaluates responses and provides feedback instantly

Progress Tracking

Detects weak topics automatically
Tracks improvement over time
Suggests areas to focus on for maximum score improvement

Aryabhata helps students move from content consumption to deliberate practice.

How we built it

The system combines content mining, statistical modeling, OCR, and large language models.

Question Mining & Test Generation

Extracted questions from JEE past papers
Mined exercises from NCERT content
Built a statistical weighting model to generate realistic mock tests aligned with exam patterns
Implemented topic-aware question selection

Answer Evaluation Pipeline

OCR pipeline for reading handwritten answer sheets
Vision + prompt-engineering workflow using Gemini
Structured evaluation prompts for scoring and feedback

Guided Solving System

Prompt-based reasoning feedback instead of solution dumping
Step-aware evaluation of student responses

Challenges we ran into

One major challenge was evaluating handwritten answers reliably, especially when solutions included:

mathematical notation
diagrams
multi-step reasoning
fractions and equations

Another challenge was designing prompts that guide students instead of giving away answers, which required careful iteration.

Generating realistic mock tests with correct topic distribution was also non-trivial and required building a weighting model from past exam papers.

Accomplishments that we're proud of

We’re proud that Aryabhata can:

Evaluate uploaded answer sheets using AI
Generate syllabus-aligned mock tests
Provide guided problem-solving feedback
Identify weak topics automatically

Most importantly, we built an experience that feels like having a personal prep coach available anytime.

What we learned

During this project, we learned:

OCR alone is insufficient for evaluating academic answers; combining image context with text improves results
Prompt design is critical for educational guidance systems
Exam-pattern-aware question generation significantly improves mock test quality
AI can support mentorship workflows, not just content delivery