Currently in development, ArkLegacy will be a comprehensive IoT analytics platform targeting mining and agricultural operations, set to transform real-time sensor data streams into structured, actionable insights for operational decision-making. The platform will combine hardware sensor integration with advanced data visualisation dashboards, enabling predictive maintenance scheduling and continuous environmental condition monitoring once deployed. The backend will process time-series data through analytics pipelines built on a Python and JavaScript stack, drawing on the data engineering principles being developed through MSc research into embedded systems and web-integrated analytics. On the frontend, the goal is to deliver an intuitive interface designed specifically for field operators, presenting complex telemetry in interpretable visual formats that require no deep technical knowledge to use. Ultimately, this project will serve as a direct application of the dissertation research focus, demonstrating how embedded systems and cloud-based analytics pipelines can support data-driven decision-making in resource-intensive industries at scale.

A complete full-stack e-commerce platform built for Royal Priesthood, a faith-inspired premium streetwear brand, developed from the ground up with a focus on performance, security, and maintainable architecture. The frontend is crafted using HTML5, CSS3, and Vanilla JavaScript with carefully selected Google Fonts (Bebas Neue, Cormorant Garamond) to reinforce the brand's premium identity, while the backend is powered by Node.js and Express.js through a RESTful API architecture. User authentication is secured with bcrypt password encryption and session management, and all product and customer data is persisted in a relational SQL database with structured query optimisation skills directly developed through my Teaching Assistant role where I facilitated data retrieval and SQL query optimisation for 100+ students. The codebase follows an MVC-style separation of concerns, with modular JavaScript files (products.js, cart.js, auth.js, main.js) ensuring scalability and straightforward onboarding for collabourating developers. This project reflects my ability to deliver commercial-grade software solutions for real clients, bridging engineering rigour with practical software development.

A Python-based Streamlit web application that automates residential electrical load profile generation using the Catalogue Method a systematic approach to estimating household energy demand by cataloguing appliance types, usage patterns, and rated power values. The application allows users to interactively define household appliance inventories and operating schedules, dynamically computing and visualising 24-hour power demand curves. Validation was performed against physically measured household loads using a Fluke power quality meter, achieving power estimation accuracy within 5–8% of real readings a commercially acceptable margin for residential solar PV sizing and demand forecasting. Built using Pandas and Matplotlib for data handling and visualisation, the tool reflects my broader expertise in data-driven engineering workflows where Python analytics pipelines translate raw measurements into actionable design insights. This project has direct applications in South Africa's renewable energy transition, providing accessible tooling for engineers and consultants designing off-grid and grid-tied PV systems for residential clients.

Developed a precision hand movement measurement system for post-stroke rehabilitation, enabling physiotherapists and patients to objectively track motor recovery progress through quantifiable distance and motion metrics. The system achieved 95% measurement accuracy with a cumulative measurement error of only 2.92% ± 0.0489%, demonstrating the reliability necessary for clinical-grade rehabilitation feedback in real-world physiotherapy settings. A key design constraint was affordability: the complete system was fabricated at a cost of R420, making it viable for deployment in resource-limited public healthcare and community rehabilitation environments where expensive commercial motion capture systems are inaccessible. The hardware architecture integrates time-of-flight and inertial sensors with embedded signal processing on a microcontroller platform, with firmware written in C to achieve the low-latency, high-accuracy measurement pipeline required. This project exemplifies the intersection of embedded systems engineering and healthcare impact, demonstrating how thoughtful hardware design can directly improve quality of life outcomes for stroke survivors in underserved communities.

Designed, simulated, and physically fabricated a directional Yagi-Uda antenna optimised for operation at 2.4 GHz, targeting Wi-Fi range extension applications in environments where standard omnidirectional routers deliver insufficient coverage. The antenna achieved a realised gain of 9 dBi, compared to the 2 dBi gain of a standard omnidirectional router antenna representing a 7 dB improvement in signal strength and a significant increase in effective communication range in the beam direction. The design process was carried out using 4-NEC2 electromagnetic simulation software to iteratively optimise element spacing, driven element length, and reflector geometry before physical fabrication, ensuring simulated performance closely matched measured results. Antenna construction used precision-cut aluminium elements mounted on a non-conductive boom, with impedance matching designed to minimise return loss at the feed point for maximum power transfer. This project reflects hands-on application of HF Techniques coursework and demonstrates the complete antenna engineering workflow from theoretical design and electromagnetic simulation through to physical fabrication, measurement, and performance validation.

Developed a Python-based GUI simulation program to model and compare the national-scale demand-side management (DSM) impact of replacing conventional electric storage water heaters with instantaneous (tankless) alternatives across South Africa's residential sector. The simulation modelled consumption behaviour across 5.4 million installations using statistical load profiles derived from real household usage data, computing aggregate demand curves and energy consumption totals under both heater configurations. Results projected a peak demand reduction of 470 MW equivalent to half a stage of Eskom load shedding alongside 32% daily energy savings amounting to 35.16 GWh per day, quantifying the national grid relief achievable through targeted appliance-level DSM interventions. Built using Python with Pandas, NumPy, and Matplotlib, the program implements a data pipeline from input parameters through statistical modelling to visualised output, making the analysis accessible and reproducible for energy policy stakeholders. This project directly addresses South Africa's ongoing energy crisis by providing evidence-based, data-driven analysis to inform government and utility-level decisions around residential appliance standards and incentive programmes.

Fabricated an intelligent embedded security system for telecommunications infrastructure cabinets, addressing the widespread theft of backup batteries from cell towers and telecoms sites a major operational and financial challenge for South African network operators. Developed during vacation work at Rain South Africa, the system implements a predictive classification model that achieved 95% accuracy in distinguishing between authorised maintenance personnel and intruders, with a detection and alert response time of under 5 seconds. The machine learning classification model was trained on sensor fusion data combining PIR motion detection, vibration sensing, and access pattern analysis, enabling the system to differentiate between routine authorised access and suspicious intrusion behaviour with high confidence. Hardware integration involved embedded firmware development in C, sensor interfacing, and a wireless alerting module that pushes real-time notifications to a remote monitoring dashboard, enabling operators to respond to intrusion events immediately. This project also involved cross-functional collabouration to produce comprehensive technical documentation and project specifications work that directly contributed to reduced engineer onboarding time, as acknowledged during the Rain South Africa vacation programme.