JudiciaryPlus is an AI-assisted study platform for judiciary and law exam preparation. A student picks their exam and syllabus, then plans, studies, practises and revises in one account, with an AI tutor that answers from legal sources rather than from memory.
Law exam preparation is spread across PDFs, notes, question papers and generic chatbots that can state the law confidently and wrongly. Students need one place that keeps their own record, focuses practice on weak areas and grounds AI answers in real statutes and judgments.
My role
Independent product. I designed and built it end to end:
React front end, serverless API, Postgres database schema and AI integration
An automated test suite of more than 1,600 tests covering scheduling, revision, AI routing, admin permissions and data migrations
Separate production and staging environments with scripted release checks
Key features
AI tutor with streamed answers, grounded in retrieved statutes and verified judgments
Adaptive daily quizzes, model papers and timed mock exams with post-exam analysis
Syllabus management and a daily study planner built within a fixed study-hour budget
Adaptive revision queue ranked by recent mistakes, accuracy and due dates
Progress dashboard, reports, study calendar and a mistake book
Admin console with server-side role checks and an append-only audit log
Parent progress reports and per-account data isolated with row-level security
Engineering highlights
Reliable AI: each feature uses an ordered fallback chain across two AI providers, with a total time limit and health tracking, so one failing provider does not leave a student stuck.
Trustworthy scoring: revision priority and mastery are calculated by fixed rules from practice results. The AI model never decides them.
Secure admin: every admin action is checked on the server against a permission list, using the role stored in the database rather than anything sent by the browser.