Back to work

CASE STUDY · EDUCATION

Edvex: eight exam tracks, one platform.

Students preparing for the SAT, GRE, MCAT, LSAT, and more were juggling a pile of disconnected apps — practice in one, flashcards in another, tutoring in a third, none of them talking to each other. Edvex replaces the pile with one platform where every feature reads from the same live picture of the student.

Client
Ariel Kopolovitz, Founder, Edvex
Industry
Education
Built by
Coder Crew

8+

Exam tracks, one codebase

13

Integrated product modules

76

Application pages

67

Automated end-to-end tests

Test prep was working against the student.

Fragmented tools, zero shared memory

Practice, flashcards, tutoring, and planning lived in separate apps. The tutor never knew what was practiced yesterday.

Static, one-size-fits-all content

Fixed question banks bored strong students and buried weak ones. Nobody got material aimed at their actual gaps.

Expert feedback was gated and slow

Essay and interview feedback meant booking a human tutor: costly, scheduled, and unavailable at 11pm before a deadline.

Guardians were flying blind

A parent's only progress signal was asking the student, which rarely produced an accurate answer.

Every new exam meant a new build

Extending from SAT to GRE or MCAT usually means a separate product, a separate pipeline, and a separate cost.

Five problems, five design answers.

Before

Fragmented tools

After

One platform, one data model. Practice, tutoring, flashcards, essays, interviews, and planning all read from the same live performance history.

Before

Static content

After

An engine that recalibrates per answer. Every answer updates the student's tier, re-targets difficulty, and triggers fresh AI-generated questions.

Before

Gated feedback

After

Grading and coaching on demand. Rubric-based essay grading and an AI interview coach respond the moment work is submitted — no booking, no waiting.

Before

Guardian blindness

After

A dedicated guardian dashboard. Analytics, insights, and forecasting go straight to the parent — no student self-reporting required.

Before

Exam-by-exam builds

After

One engine, eight tracks. Shared adaptive infrastructure tuned per exam shipped SAT through Bar from a single codebase.

A learning loop that never stops recalibrating.

Most platforms serve a fixed question bank. Edvex treats every single answer as a signal — and rebuilds the student's next question around it.

1

Answer evaluated

Correctness, timing, and topic captured

2

Tier updated

Performance tier recalculated live

3

Difficulty reset

Next-question difficulty re-targeted

4

AI generates

A fresh question is written, duplicates blocked

5

Delivered

The student sees material tuned to right now

loop

Study packs and the review system run on the same signal: packs assemble questions at the right difficulty, and review resurfaces only what the engine has flagged as weak.

Thirteen modules, one data spine.

Every module reads and writes the same performance history — a flashcard reviewed or an essay graded sharpens the same picture the adaptive loop and guardian insights run on.

Student dashboard

Progress, sessions, and metrics in one tier-aware view

Adaptive practice

Every question tuned to the student's live tier

Exam simulation

Timed, true-to-format mock exams

Smart review

Resurfaces only flagged weak areas, not a generic re-test

AI tutor "Ed"

Voice-enabled tutoring with multi-session memory

Materials processing

PDFs and notes auto-become summaries, terms, and questions

Flashcards

Spaced repetition scheduled per card for retention

Study planner

AI schedules study phases around real availability

Essay grading

Rubric-based scoring with concrete feedback over time

Interview practice

Behavioral and technical modes with personas and debriefs

Community

Forums, study groups, and shared study packs

Guardian dashboard

Analytics, forecasting, and secure messaging for parents

Admin control center

Subscriptions, AI quality review, moderation, system health

What shipped.

  • A stack of separate tools replaced by one cross-linked product

  • Static question banks became continuous per-answer recalibration

  • Guardians see real progress without student self-reporting

  • Eight-plus exam tracks shipped from a single codebase

  • Sixty-seven automated end-to-end tests validated every track before launch

How it's built.

FrontendReact 18 · Vite · Tailwind CSS · Radix UI · Framer Motion
BackendSupabase: PostgreSQL, Auth, Storage, Realtime
AIOpenAI GPT-4o and GPT-4o-mini · Whisper · text-to-speech
Serverless35 Supabase Edge Functions (Deno); keys never touch the client
PaymentsStripe Checkout, subscriptions, webhooks, 7 plans
SecurityRow-level security across 20+ entities: student, guardian, and admin isolation
TestingPlaywright: 67 automated end-to-end tests across all exam tracks

Why Coder Crew.

One accountable team delivered the architecture, the AI integration, the security model, and the test suite — not a loose assembly of contractors.

Have a build that needs shipping — or a business still running by hand?

Either way, we start the same place: understand first, then build. The audit is one week, one fixed price, credited toward the work.