I'm Badr Aqajjif — studying Réseaux Informatiques at EST Khénifra, and spending the rest of my time designing, deploying and keeping alive backend services, security tooling and cloud infrastructure that run in production, not just in a repository.
Before EST Khénifra, I was already writing Python scripts, automating servers with Bash, and building small apps in Flutter and Node.js — mostly by breaking things on my own Linux boxes until they worked. The DUT in Réseaux Informatiques gave that a name and a structure: networking, systems administration and security, taught properly.
Outside coursework, I run a small personal infrastructure stack on an Oracle Cloud VPS — streaming platforms, a family communications app, and a growing set of network security experiments. It's less about any single project and more about keeping real systems alive: watching logs, chasing down a silent connection drop, hardening an endpoint after a bad night of traffic.
Python scripting, Bash automation, Flutter/Dart, Node.js and Linux server administration — learned by building, before any formal program.
DUT coursework in networking, systems and security, alongside a growing set of personal infrastructure projects.
An Oracle Cloud VPS running FastAPI services, Nginx, MySQL/MariaDB and Dockerized tooling — administered end to end.
Hybrid Suricata + machine-learning intrusion detection, FreeRADIUS deployment, and applied network defense.
A React + Flask forensics platform for steganography detection, built as my end-of-program project (PFE).
Not mockups — this is the actual infrastructure map behind my personal projects: services I've built, deployed and am still on call for. Click a node for details.
Streaming platform core
Three-component streaming system — a FastAPI backend with a Telegram bot admin interface, plus separate Flutter apps for users and admins. JWT auth, certificate pinning and device binding on the client side.
REST & bot API layer
Token-based auth over HTTPS with a sliding-window rate limiter guarding the verification endpoint against brute-force attempts — a deliberate trade-off over signature-based schemes.
Realtime comms application
A family walkie-talkie app — bidirectional audio over LiveKit, push-triggered wake via Firebase Cloud Messaging, and a Node.js signaling backend.
Cybersecurity sandbox
A hybrid intrusion-detection sandbox pairing Suricata signatures with an Isolation Forest anomaly model, plus a Random Forest + XGBoost classifier trained on CIC-IDS2017, surfaced through a Flask dashboard.
Persistence layer
MySQL/MariaDB backing the streaming and bot platforms, with a SQLite-backed FreeRADIUS deployment on the same VPS — a workaround for an ARM64 MariaDB incompatibility.
Infrastructure node
The Ubuntu host underneath everything above — Nginx as reverse proxy, PM2-managed Node services, Dockerized Nextcloud, and a backup/failover node in planning.
Four areas I keep coming back to — in coursework and outside it.
Systems I designed, built and — in most cases — still operate.
A complete streaming ecosystem: a FastAPI backend driven partly through a Telegram bot admin interface, plus dedicated Flutter apps for end users and administrators. Handles auth, device binding, ban enforcement and live broadcast/announcement systems.
A family walkie-talkie app built in Flutter, using LiveKit for bidirectional real-time audio and Firebase Cloud Messaging for wake-on-call notifications, backed by a Node.js signaling service.
A hybrid intrusion-detection layer combining Suricata signature rules with an Isolation Forest anomaly model, and a Random Forest + XGBoost classifier trained on CIC-IDS2017 — automatically blocking DDoS, brute-force and SSH-level attacks via ufw.
A Bash-orchestrated pipeline that monitors live streamers, auto-creates YouTube broadcasts through the Data API v3, and pipes video through FFmpeg — with a background OAuth refresh worker and named-pipe handling for reliable, long-running streams.
A React + Flask web platform for steganography forensics: multi-format detection across LSB, chi-square, echo-hiding and spectral analysis, with automated PDF reporting and MD5/SHA-256 evidence hashing.
A self-hosted messaging app built for familiar, WhatsApp-like UX — real-time messaging over WebSockets, WebRTC voice/video calls, voice notes with waveform playback, and an emergency-wipe feature that clears a conversation for both ends at once.
Utilities I put together for specific problems — smaller in scope, still running.
Telegram bot built against TikTok's unofficial mobile API, with device fingerprinting matched to a real handset profile for reliable responses.
A desktop downloader reworked into a Telegram bot with python-telegram-bot and yt-dlp — same functionality, no install required.
Converts landscape video to 9:16 with FFmpeg on request, running as a systemd service on an Ubuntu VPS.
Open to internship opportunities, collaboration on network & security projects, or just a conversation about infrastructure. Reach out through whichever channel is easiest.