EduAgent

Inspiration

SAT prep tools tend to either throw endless generic practice questions at students or lock personalized coaching behind expensive tutors. I wanted to build something that actually watches where a student is struggling — down to specific sub-skills like "Quadratic Functions" or "Data Interpretation" — and adapts in real time, the way a great human tutor would.

What it does

EduAgent AI is an adaptive SAT prep platform that:

• Tracks mastery across subjects (Math, Reading & Writing, Algebra, Data Analysis, Geometry) with live percentage breakdowns
• Surfaces priority gaps using a spaced-repetition-style system, flagging exactly which skills need review and when
• Projects a live SAT score estimate based on performance trends
• Offers an AI Tutor for on-demand explanations and a Study Room for guided sessions
• Generates a daily study plan mixing adaptive practice, timed reading passages, and combined-subject review

How I built it

The frontend is built with Vite for fast iteration, paired with a Node.js backend. For the AI tutor and question-generation layer, I integrate with Groq's API (using llama-3.3-70b-versatile) for low-latency responses at hackathon-friendly cost. The dashboard renders mastery data, gap analysis, and study plans dynamically based on the student's practice history.

Challenges I ran into

• Migrating my AI integration from the Anthropic API to Groq mid-build, to balance response speed and cost for a live demo
• Designing a gap-detection system that's meaningful (not just "lowest score") — weighting recency, question difficulty, and spaced-repetition due dates
• Keeping the UI dense with useful data (mastery bars, gap cards, study plan) without it feeling cluttered

What I learned

• How to structure prompts for consistent, tutor-quality explanations from an LLM
• Practical trade-offs between AI providers (cost, speed, quality) for a real-time educational product
• The value of surfacing actionable next steps (a "Start Gap Practice Session" button) rather than just displaying analytics

What's next for EduAgent AI

• Expanding subject coverage beyond SAT Math and Reading & Writing
• Adding a parent/teacher view for progress reporting
• Building out spaced-repetition scheduling more rigorously across all skill nodes

📌 Demo Data Notice

This deployed demo (https://eduagent-ai-mu.vercel.app/) is pre-seeded with sample student data — mastery scores, question history, streaks, and achievements — so reviewers can explore the full feature set immediately without needing to answer 100+ questions first.

In a production environment, all of this data (subject mastery %, skill gaps, SAT projection, XP/badges) is generated live as a real student interacts with the platform — it is not hardcoded or faked in the underlying logic.

Built With

• express.js
• groq
• node.js
• react
• vercel
• vite