PrepAI

AI-powered SAT prep inspired by Khan Academy, using real College Board questions and generating unlimited SAT-style problems tailored to your strengths.

Comment

Inspiration I built PrepAI because I wanted to help my younger sister, who is currently preparing for the SAT, study more effectively. When I was in high school three years ago, I had access to Khan Academy’s SAT prep program, which had plenty of free resources. However, the new digital SAT no longer offers the same amount of free material. Students now only have access to a few practice tests through Bluebook compared to the dozens available for the old version. I also know how expensive SAT prep can be, with tutoring, Kaplan, and Princeton Review books costing hundreds of dollars. I wanted to create a free, modern, and AI-driven platform that could give every student access to unlimited SAT-style practice.

What it does PrepAI is an AI-powered SAT preparation platform that provides students with personalized practice. It pulls authentic College Board SAT questions and reformats them into interactive quiz-style problems instead of full test sets. Each student’s accuracy is tracked by subject, skill area, and difficulty to identify weak spots and recommend targeted practice. Using AI, the app can also generate new, realistic SAT-style questions, creating an unlimited study experience.

How we built it I built the backend using Python (Flask) and integrated Firebase/Firestore for user data management and question storage. The frontend was created with React, giving users a smooth and minimalistic testing experience similar to Bluebook. I used the Google Gemini API to generate new SAT-style questions modeled after real College Board problems. For data, I extracted and reformatted real SAT questions into a modular quiz format that allows users to focus on one skill or topic at a time. I also implemented performance tracking so that each user’s accuracy and progress are logged and used to recommend specific practice areas.

Challenges I ran into One of the biggest challenges was getting AI to consistently generate SAT-style questions with correct formatting and rationales. Cleaning and structuring thousands of College Board questions into usable data also took time and attention to detail. Handling longer AI generation times without freezing the frontend required optimization and asynchronous request handling. Another challenge was balancing accurate question content with a user-friendly interface that feels intuitive for students.

Accomplishments that I am proud of I am proud that PrepAI gives students free access to high-quality SAT-style questions while providing personalized feedback. I am also proud of integrating Gemini successfully to generate new problems that mimic official SAT content. Seeing the app suggest practice areas based on user accuracy felt like a breakthrough in adaptive learning. Most importantly, I am proud that this project started as a way to help my sister and turned into a tool that can help many students prepare effectively for the SAT.

What I learned I learned how to design adaptive learning tools that use both AI and user data to improve the study experience. I gained valuable experience building prompts for consistent AI generation, structuring backend APIs, and tracking analytics with Firebase. I also learned how to integrate multiple technologies together in a scalable architecture while maintaining performance and usability.

What's next for PrepAI I am currently working on fine-tuning AI models like OpenAI and Gemini to generate more accurate and diverse SAT-style problems. I am also deploying the full application on Google Cloud Platform (GCP) for better scalability, performance, and integration with other Google services. In the future, I plan to add full-length adaptive practice tests, peer study features, and progress analytics dashboards so students can see how their scores improve over time.

Built With

Share this project:

Updates