Imora
IELTS prep that marks writing and speaking the way the exam does — then keeps you on a plan until test day.
Most IELTS products sell a pile of PDFs and a score. Students still do not know which sentence cost them the band, or what to study tomorrow.
Imora is the product I am building now at Imora.uz. It puts real Cambridge papers on a test-day screen, marks writing and speaking against the official criteria, and rebuilds the daily plan when someone misses a day.
Role
I am the full-stack engineer on the learner product. I own the practice flows, how scoring comes back to the student, and the planner that turns a target band and a test date into today's work.
Constraint
Prep and admissions are usually two vendors. The student pays for mocks in one place, then pays an agency again, and neither side has a record of what they actually did. Imora has to be both — exam practice and the path after the band — without feeling like two apps glued together.
The other constraint is honesty. A writing score that cannot point at the phrase that hurt lexical resource is just another number. Feedback has to be specific enough to act on.
What I built
- Exam-faithful practice. Full mocks and drills cut by question type, on a screen that matches how the computer-based test feels.
- AI marking on all four skills. Writing and speaking come back scored on the official criteria, with the lines that cost marks rewritten.
- Traceable reading and listening. A wrong answer should show the sentence or the second of audio that decided it — not only the letter you missed.
- A plan that moves. Give it a test date and a target band. Miss a day and it re-plans. The habit layer (XP, streaks, levels) is there because three months of showing up is the actual product.
Stack
Next.js, TypeScript, and the rest of a modern full-stack app. The interesting part is not the framework list — it is wiring model output to official IELTS descriptors so a student can trust the mark.
What I would do differently
I would have instrumented "did this comment change the next essay" from week one. Scoring is only useful if the next attempt is better. That loop is what I am tightening now.