I'm a Mechatronic Engineering graduate from Universiti Sains Malaysia (USM) who builds where hardware, software, data, and intelligence meet. My work moves between real-time firmware, signal processing, on-device machine learning, connected devices, manufacturing analytics, automation, and native application development.
I care about the full engineering path: understanding the physical signal or production data, choosing a practical architecture, working within real constraints, and leaving behind a system that can be tested, explained, and improved.
|
A requirements-based command-line toolkit that turns a validation plan into bounded workloads, privacy-conscious inventory, immutable evidence, readable reports, and compatibility-gated regression comparisons. The portfolio MVP now covers the full workflow: capability discovery, CPU, memory, and temporary-storage validation, telemetry, explicit requirement evaluation, SQLite persistence, JSON/Markdown/HTML reports, known-good baselines, conservative comparison policies, and a clearly labelled fault-injection demonstration.
|
Portfolio MVP implemented ✓ 66 automated tests passing Active development · 0.1.0.dev0 |
The original development machine still lacks the optional CMake/C++20 toolchain. The native decoder is instead built and tested by the passing GitHub Actions matrix on both Windows and Ubuntu; unavailable local capabilities remain explicit WARN or SKIP evidence.
|
An end-to-end production-test analytics system that turns 8,000 deliberately messy tester records into validated manufacturing insights and an interactive Power BI dashboard. PostgreSQL preserves the raw source, cleans and quarantines records, enforces a normalized production model, and exposes reusable analytics views for yield, Pareto, station performance, and trend analysis.
|
| Test records | First-pass yield | Analytics contract | PBIR validation |
|---|---|---|---|
| 8,000 | 94.48% | 4 reusable views | 0 errors |
|
A complete embedded Edge AI pipeline that recognizes six domestic sound categories directly on a Renesas RA8P1 Titan Board—without cloud inference. The firmware captures microphone audio, extracts MFCC, delta, and delta-delta features, runs an INT8 convolutional neural network through TensorFlow Lite for Microcontrollers, and produces confidence-gated alerts in real time.
|
|
| Sound classes | Feature pipeline | Model footprint | Board-recorded evaluation |
|---|---|---|---|
| 6 | 39 × 61 MFCC-derived features | 114.4 KB INT8 model | 87.5% on a small test split |
|
MyBudget is a native Windows monthly budget planner designed around private, PC-local data. It handles planning, transactions, recurring income and bills, savings goals, investments, reporting, backup, and CSV exchange. The codebase separates UI, budget rules, and persistence; uses exact decimal money calculations; and includes data-preserving SQLite migrations plus automated verification.
|
| Native screens | Automated tests | Data model | Cloud dependency |
|---|---|---|---|
| 8 | 105 | Local SQLite | None |
| Domain | What I build with |
|---|---|
| Embedded systems | C, C++, microcontrollers, RTOS concepts, firmware architecture, peripheral and sensor integration |
| Edge AI & signal processing | TensorFlow Lite Micro, TinyML, CNNs, MFCC feature extraction, CMSIS-DSP, quantized inference |
| Manufacturing analytics | PostgreSQL, advanced SQL, Power BI, data modeling, quality validation, yield and failure analysis |
| Platform validation | Python, C++20, bounded workloads, telemetry, requirements evaluation, immutable reports, regression baselines, cross-platform CI |
| Automation & control | PLC Ladder Logic, control systems, MATLAB, Simulink |
| Connected devices | MQTT, UART, LTE AT commands, Wi-Fi, OTA update workflows |
| Application software | C#, .NET, WinUI 3, XAML, MVVM, SQLite, Python |
|
Innowave LLC
|
Universiti Sains Malaysia A multidisciplinary foundation across electronics, embedded programming, control, automation, mechanical systems, and intelligent system design. That mix still shapes how I work: system-first, evidence-led, and comfortable crossing hardware–software boundaries. |
| System | Engineering focus |
|---|---|
| Legged line-following robot | Arduino sensing, locomotion, and closed-loop path following |
| Hand-hygiene monitoring system | PLC-controlled automation and process monitoring |
| Height-measurement device | RISC-V and 8052 embedded implementation |
| FPGA passcode alarm | Digital logic, state-based control, and hardware implementation |
- Extending a verified PC platform validation MVP with reproducible plans, immutable evidence, offline reports, and regression baselines.
- Building production-minded embedded and Edge AI systems.
- Applying SQL and Power BI to manufacturing yield, failure, and process data.
- Strengthening real-time firmware architecture, testing, and documentation.
- Expanding into well-structured application software and AI-assisted engineering workflows.
- Publishing project evidence—not just finished screenshots, but architecture, constraints, verification, and lessons learned.
I'm open to graduate and entry-level opportunities across embedded systems, Edge AI, manufacturing analytics, automation, IoT, and adjacent software engineering.
Email · LinkedIn · Repositories
Sense precisely. Model clearly. Decide locally. Build reliably.
