Arbitrary Programmers

Sign in Page
Sign Up Page
Quiz Page
Uploaded Notes Page
Create a study plan through the plane ticket (Second Half)
AI Summary for the materials that the users have uploaded
Flashcard Page
Quiz History Page
Create a study plan through the plane ticket (First Half)
Dashboard (Lower Part)
Dashboard (Upper Part)
Fligh ticket place where users can click complete after they finish they study plan
AI explaining the answer to different age group
Sepolia Cryptocurrecny Transaction
AI Companion
Badge Page for users and see the NFT they get

Inspiration

We were inspired by a common student pain point: students spend too much time organizing lecture materials and not enough time actually learning. As Arbitrary Programmers, we wanted to build a system that converts raw lecture content into active study workflows while keeping learners motivated through visible progress and rewards.

Our guiding principle was:

$$ \text{Effective Learning} \propto \frac{\text{Practice} \times \text{Consistency}}{\text{Friction}} $$

So our goal became clear: increase practice and consistency while reducing friction.

What it does

ScholarFly allows users to upload lecture content (notes, audio, or video) and automatically generates:

Transcripts
AI summaries
Quizzes
Flashcards
Flight-ticket checkpoints

It also introduces gamification and Web3 rewards:

Complete a ticket $\rightarrow$ mint an NFT badge
Receive SFT token rewards, which could be used for discounting on models

How we built it

We built ScholarFly using:

Next.js + React + Tailwind CSS for frontend
Next.js API routes for backend logic
PostgreSQL for authentication, database, and storage
Azure OpenAI + AssemblyAI for AI generation and transcription
Ethers + Solidity smart contracts on Sepolia for NFT and ERC-20 rewards

We designed a shared authenticated app shell and connected wallet-based reward flows directly to learning milestones.

Challenges we ran into

Hydration and client-render mismatches in dynamic UI components
Stabilising AI output quality across different ages and learning styles
Making NFT metadata reliably visible after mint (token ID/log parsing, metadata handling)
Managing wallet/network constraints (Sepolia switching, RPC/env setup, minter permissions)
Keeping UX consistent while shipping many features quickly

Accomplishments that we're proud of

Built an end-to-end pipeline: upload lecture $\rightarrow$ generate study assets $\rightarrow$ reward completion
Delivered a reusable ticket/checkpoint system with session history and clickable flows
Implemented live badge + token reward mechanics tied to real learning completion
Added global SFT balance visibility across authenticated experiences
Maintained coherent product identity and UX through rapid iterations

What we learned

Prompt engineering can matter as much as model selection for educational quality
Small UX details (loading states, clear statuses, simple navigation) strongly improve user trust
Web3 integration reliability depends on operational details, not only contract code
Modular architecture (shared utilities, shared shell, clear API boundaries) speeds iteration