PentaAI — Exam Evaluation and Guidance Platform

PentaAI automates the job that costs exam-running institutions the most time: connecting the publisher's answer key, the optical reader's output file and the student list.

The institution uploads the publisher's learning-outcome spreadsheet; the system links every question to its subject, topic and outcome. When the optical file is uploaded, students are matched by number and name, and net scores, points and rankings are calculated. Each student gets a branded PDF report card with question-level analysis and a list of weak topics.

On the guidance side the system looks at trends rather than single exams. Gaps that repeat across several exams land in the Outcome Tracking Centre as cards. PentaRank combines exam performance and study consistency into one score; the dashboard shows on the first screen which students are rising and which are slipping.

The architecture is multi-tenant from day one. Every institution signs in from its own subdomain, and isolation is not left to the application layer alone: with PostgreSQL Row Level Security, one institution's students, exams and payments are invisible to another's queries at the database level. Student, parent, teacher and admin sessions are kept apart, and each institution decides which modules its guidance teachers can see.

The stack is a Next.js 15 and NestJS monorepo. PDF report cards and bulk jobs run in the background on a Redis-backed queue, and files live in S3-compatible storage. The platform is live and used by more than 1,000 students.