Skip to content
English, current language Français
  • Full-Stack Applications
  • Embedded & Systems Software
  • DevOps & Cybersecurity
  • UI/UX & Product Design

SecureExam

Independently designed and developed: an end-to-end platform to prepare, deploy and supervise computer-based exams in secured environments.

Independently designed and developed SecureExam, an end-to-end examination management and secure environment orchestration platform. Final-year engineering project (Bureau d’études), ISEN Méditerranée, 2025–2026, about three months, with academic guidance from ISEN. The project was validated academically.

Overview

SecureExam lets an educational institution run exams on computers while controlling what students can use during the exam: which software, which websites, which system permissions. Professors prepare the exam environment, an agent applies it on each machine, and the platform collects the students’ work at the end.

Context and challenge

A computer-based exam is only fair if every student works in the same, controlled environment. Preparing machines by hand for each exam is slow and error-prone, and it is hard to check during the exam that a machine still matches what was planned. The challenge: turn an exam’s rules into a configuration that can be applied to many machines, followed live, and removed afterwards.

Objectives

  • Let professors describe an exam environment without system administration skills.
  • Apply it to student machines automatically.
  • Follow sessions while the exam runs.
  • Collect every student’s work reliably at the end.
  • Manage users, machines and configurations in one place.

My role

I carried out the whole project on my own: architecture and technical design; UI/UX design in Figma; the Angular and TypeScript front end; the Python and FastAPI back end, with its SQLite database; authentication and security features; the Go agent on NixOS and its integration with the exam machines; Docker deployment; testing, integration and documentation. ISEN supervised the project academically.

Architecture

Three parts, which talk to each other through REST APIs. The services run in Docker containers, behind Nginx.

Web application

Spaces for professors, students and administrators.

Angular · TypeScript

Back end

Users, exams, configurations, machines and submissions; authentication.

Python · FastAPI · SQLite · JWT

Exam agent

On each exam machine: prepares it, applies the configuration, follows the session, returns the work.

Go · NixOS

Three spaces

Professors

Choose a predefined exam environment or build their own; select the authorised software, websites and domains, and system permissions; assign students and machines; follow the session; retrieve the submissions.

Students

See their assigned exams, start their session, work in the prepared environment and submit their final work.

Administrators

Manage users, machines and exam configurations, monitor the platform’s resources and handle operations. A support interface completes the platform.

The exam agent

The agent turns a configuration into reality on each machine. It:

  1. retrieves the exam’s configuration from the back end;
  2. prepares the machine;
  3. applies the restrictions: software, network and domain access, and system permissions;
  4. follows the session;
  5. collects the student’s work and sends the final submission.

NixOS describes a machine’s whole configuration declaratively, so an exam environment can be applied, and removed, in a reproducible way.

Key capabilities

  • Reusable exam environments, predefined or custom.
  • Per-exam control of software, network access and permissions.
  • Assignment of students to machines.
  • Session follow-up during the exam.
  • Automatic collection of submissions.
  • Distinct roles and permissions for professors, students and administrators.

UI/UX

I designed the interface in Figma before building it. Each space shows only what its users need: professors build an exam step by step; students see one clear action at a time; administrators get an overview of users, machines and resources.

Engineering challenges

  • One description, many machines: turning a professor’s choices into a configuration that the agent can apply, then reverse.
  • Coordination: keeping the back end and every agent in agreement on each machine’s state, before, during and after the exam.
  • No lost work: collecting submissions reliably at the end of a session.
  • Security by roles: JWT authentication and permissions specific to each space.
All projects

Get in touch

Interested in a project like this one?

Tell me about your project or role.

Discuss a project