[RETURN TO CORE]

Architecture & Projects.

An in-depth technical breakdown of end-to-end environmental pipelines, cybersecurity EDR designs, deep learning models, and autonomous hardware systems.

Smart Eco-Monitoring Platform (SEMP)

Capstone // Humber Polytechnic
Sep 2025 – Apr 2026
[SYS_ACTIVE_DEPLOYED]
SEMP is an end-to-end environmental monitoring system, from battery-powered field sensors buried in the Humber Arboretum's soil, to a live web dashboard that alerts staff to critical conditions in real time and uses machine learning to detect plant disease from uploaded leaf images.

Architecture Built

  • Sensors: 6 Dragino LoRaWAN devices measuring soil moisture, pH, temperature, humidity, water depth, and flow (5–10 yr battery).
  • Data pipeline: Wireless transmission via The Things Network into a Supabase PostgreSQL cloud DB (2.1s latency).
  • Dashboard: React.js web app on Vercel showing live readings, historical trends, weather, and thresholds.

ML Plant Health Model

  • Trained custom SimpleCNN vs. ResNet18 transfer learning in PyTorch on a large multi-class dataset with lighting augmentation.
  • Selected SimpleCNN for low-resource hardware efficiency, yielding 85–92% field accuracy.
  • Results: 97.3% packet delivery, 0 packets lost in 48h stress test. Deployed for $1,431 CAD (under budget).
PythonPyTorchLoRaWANReact.jsPostgreSQLMachine Learning

Secure Network & Block Cipher Exploitation

Cryptography // Humber Polytechnic
Feb 2026 – Apr 2026
[SEC_REDESIGNED]
A hands-on cryptography project replicating the real-world BEAST (Browser Exploit Against SSL/TLS) vulnerability in a containerized lab environment, covering attack execution, mathematical cryptanalysis, and secure redesign.

Technical Scope & Exploitation

  • Docker Lab: Three isolated containers (attacker, server, client) on a virtual bridge network.
  • Oracle Server: C++ vulnerable AES-128-CBC server deliberately mimicking TLS 1.0 predictable IV flaws.
  • CPA Attack: Built a Python/Scapy script that predicts IVs, crafts XOR payloads, and successfully recovers encrypted secrets.

Cryptanalysis & Mitigation

  • Derived the algebraic XOR proof showing how predictable IVs bypass AES-128 entirely.
  • Evaluated CBC, CTR, and CFB modes against parallelization and error isolation metrics.
  • Redesign: Migrated to CTR mode (immune to IV prediction). Validated via a 7-test automated suite ensuring the BEAST attack fails.
CryptographyDockerScapyC++Network Security

Social Engineering & EDR Design

Red/Blue Team // Humber Polytechnic
Feb 2026 – Apr 2026
[EDR_HARDENED]
A cybersecurity simulation covering the Weaponization, Delivery, and Installation kill chain phases—crafting evasive payloads, tracing forensics, and designing layered Endpoint Detection & Response (EDR) controls.

Attack Vectors & Forensics

  • Payloads: VBA macro reverse shell (.doc), SEToolkit HTA executing via mshta.exe, and Base64 obfuscated PowerShell.
  • Forensics: Used Sysinternals TCPView to trace WINWORD.EXE network connections.
  • Used Process Explorer to trace malicious parent-child chains (WINWORD.EXE → cmd.exe).

EDR Architecture (4 Layers)

  • GPO: Disabled Office macros and blocked legacy .doc formats.
  • AppLocker: Deny rules applied to mshta.exe, wscript.exe, cscript.exe.
  • Event Logging: PowerShell Script Block Logging (ID 4104) to reveal decoded Base64 commands.
  • Auditing: Process Creation Auditing (ID 4688) alerting on Office spawning shells.
Penetration TestingKali LinuxSysinternalsEDR PolicyGPO

VyBe — Travel & Discovery App

Full-Stack Mobile // Humber Polytechnic
Mar 2024 – Apr 2024
[PROTOTYPE_APK]
A full-stack cross-platform mobile travel companion built in Flutter, allowing users to search any city and instantly get local destination recommendations and real-time weather forecasts.

Application Features

  • Firebase Auth registration with input validation and a radio widget toggle.
  • Live weather widget pulling real-time temperature, humidity, and 5-day forecasts.
  • World Destinations search via Places API for local landmarks, ratings, and imagery.
  • Persistent settings (Dark Mode, Airplane Mode) utilizing SwitchListTile and SharedPreferences.

Technical Implementation

  • 6+ stateful widgets with multiple state variables for dynamic UI rendering.
  • Firebase Firestore for NoSQL user data persistence across multiple app contexts.
  • UI Animations implemented via AnimatedContainer, AnimatedCrossFade, and AnimatedOpacity.
  • Compiled and deployed as a standalone Android Package Kit (APK).
FlutterDartFirebaseREST APIs

Smart Room Monitoring System

IoT & Deep Learning // Humber Polytechnic
Sep 2023 – Dec 2023
[IoT_CLASSIFIER]
A facial recognition security system for an IoT smart room, classifying individuals as authorized or unauthorized in real time using a Deep Neural Network tailored for low-resolution embedded hardware.

Neural Network Pipeline

  • Curated dataset with 80/10/20 split, preprocessed for IoT-grade resolution limits.
  • Sequential DNN in TensorFlow/Keras with hidden dense layers, binary cross-entropy loss, and Adam optimizer.
  • Data augmentation (RandomFlip, RandomRotation, RandomZoom) to reduce overfitting.

Deployment & Security

  • Live inference engine built with OpenCV for real-time webcam capture and confidence scoring.
  • Defined IoT security protocols: IAM access controls, data encryption at rest/transit.
  • Designed network segmentation architectures for IoT hardware isolation.
PythonTensorFlowOpenCVDeep Learning

Full-Stack Tic-Tac-Toe AI System

Cross-Course Capstone // Humber Polytechnic
Jan 2023 – Apr 2023
[MINIMAX_AI]
An integrated AI game system spanning four technical pillars: Python game logic, MySQL database management, Google Cloud networking, and Raspberry Pi embedded hardware.

Engine & Database

  • Python engine featuring Human, Random, and Smart AI modes using the Minimax algorithm for optimal move selection.
  • Normalized MySQL schema (Player, Game, Move, Settings) utilizing AES encryption for credentials.
  • RBAC implementation restricting SQL execution privileges.

Hardware & Cloud

  • Raspberry Pi + SenseHAT configured as physical interface (joystick input, RGB LED board display).
  • Google Cloud Platform integrated via Sheets API/OAuth.
  • Real-time telemetry feeds match data natively into a cloud-hosted Google Sheet scoreboard.
PythonMySQLRaspberry PiGCP Cloud

4WD Car Maze Navigation

Embedded Systems // Humber Polytechnic
Jan 2022 – Apr 2022
[PROTOTYPE]
A mechanical and control logic design project building a functional autonomous car capable of Bluetooth-controlled navigation and real-time ultrasonic obstacle avoidance.
  • • Integrated Arduino microcontrollers, ultrasonic sensors, and Bluetooth modules onto a 4WD chassis.
  • • Programmed motor control and collision avoidance logic natively in C.
  • • Utilized SolidWorks to model car geometry and validate spatial clearance tolerances.
ArduinoC LanguageSOLIDWORKSSensors

Iterative Marble Track Coaster

Engineering Design // Humber Polytechnic
Sep 2021 – Dec 2021
[PHYSICS_MODEL]
A first-year foundational project applying principles of mechanics, sustainability, and iterative prototyping to construct a functional coaster within strict velocity and timing parameters.
  • • Engineered track curvature utilizing mathematical principles of gravity and kinetic energy transfer.
  • • Selected recycled plastics to fulfill rigorous cost-constraint and sustainability requirements.
  • • Executed formal technical reporting validating design tradeoffs.
Engineering DesignPhysicsPrototyping