JPT Level-Up · Japanese test prep app
JPT Level-Up is our own study service for the JPT (Japanese Proficiency Test), built and run in-house from planning onward. Each answer updates both the learner's level and the question's difficulty, so every day brings five questions from the weakest part, and words from missed questions go straight into a vocabulary list. Progress is shared across desktop and mobile (installed to the home screen) under one account.
Study flow
Before
Practice, error review and vocabulary each kept in a separate place
After
Practice → explanation → error notes → vocabulary in one record, with the next question chosen automatically from the weakest part
To protect client information, these screens recreate the same features with mock data.
Background
Test prep usually means a workbook, an error notebook and a vocabulary list kept separately. Weak areas are picked by feel, and missed words are copied out again by hand. Practice on a phone during the commute and on a computer at home ends up in two separate records.
What we built
- Adaptive questions — Each answer updates the learner’s level (by section and part) and the question’s difficulty. The daily set is five questions from the weakest part, each with roughly a 70% chance of a correct answer. The expected score is shown as a range, not a single number, and appears only once a section has 30 answered questions
- Option-by-option explanations and automatic vocabulary — Every option explains why it is right or wrong. Words from missed questions go into the vocabulary list and come back on a spaced repetition schedule (FSRS)
- Web and mobile — One account record carries across desktop and mobile. On iPhone it can be added to the home screen and used like an app, with a bottom tab bar and an offline screen. The interface is available in 한국어 · English · 日本語

- Question factory — A question goes live only after generation → automated checks → three blind re-solves with the answer hidden → critique → instructor review. Operators run and review the same process from the admin screen

Results
Launched in November 2026 and in operation. Listening audio is synthesized in advance on our own GPU server, and the service is deployed on a single server without downtime.