Mechatronic Engineering graduate · Universiti Sains Malaysia
Building where physical systems, software and intelligence meet.

LinkedIn  ·  All repositories

I'm Fawwaz, a Mechatronic Engineering graduate from USM. I work across microcontroller firmware, signal processing, on-device machine learning and practical engineering software. I enjoy connecting the pieces: a physical signal, a constrained device, a useful interface—and a clear way to test the result.

My final-year acoustic detection project is the starting point. The projects below extend that foundation into platform validation, manufacturing analytics and native Windows applications.

Listening at the edge. Final-year project · Embedded Edge AI Recognising six domestic sound categories directly on a Renesas RA8P1 Titan Board, without cloud inference. Microphone capture → MFCC, delta and delta-delta features → INT8 CNN → confidence-gated alerts. An application-layer firmware snapshot connects signal processing with on-device inference and a later LCD extension. C/C++ RT-Thread CMSIS-DSP TFLM Explore the FYP → Core firmware · Evaluation figure

THE FLAGSHIP / SIGNAL → DECISION

_{Embedded prototype. The small historical split demonstrates feasibility, not broad reliability. The model-quantization correction is merged: 16 host tests and a fresh local firmware build pass. Software verification is documented; the corrected revision has not been tested on the board.}

ACTIVE BUILD / REQUIREMENTS → EVIDENCE

Test the system. Keep the evidence. Functional MVP · Continued development A requirements-based CLI for bounded CPU, memory and temporary-storage checks, capability discovery and telemetry. Preserves runs in SQLite and JSON, Markdown and HTML reports. Compatible baselines support regression comparisons; a labelled fault-injection demo exercises the failure path. Python C++20 pytest SQLite Inspect the toolkit → Verification scope · Paired runs · Failure demo

_{The native CPU probe is locally built and validated; temperature telemetry remains unavailable. Two separate quick runs preserve real timing variation and comparison threshold crossings. These are functional captures, not hardware certification or a benchmark ranking.}

From tester data to engineering insight. SQL + Power BI · Synthetic-data demonstration Turn 8,000 deliberately messy synthetic tester records into a validated PostgreSQL model and an interactive Power BI dashboard. Preserve the raw source, clean and quarantine records, then expose reusable views for yield trends, failure analysis and station performance. PostgreSQL SQL Power BI Python Explore the analysis → Dashboard · Data verification

DATA SYSTEMS / RAW RECORDS → INSIGHT

_{Generated scenarios, not production-factory results. Dashboard images are real Power BI captures of this synthetic fixture.}

SUPPORTING PROJECT / LOGIC → EXPERIENCE

MyBudget. Your data stays local. Native Windows application · Released A monthly budget planner with exact decimal calculations and local SQLite storage. Covers planning, transactions, recurring income, goals and reports. Separates the interface, domain rules and persistence, with automated tests and data-preserving migrations. C# .NET WinUI 3 SQLite Explore MyBudget → Release · Verification

_{Source-available under PolyForm Strict; unsigned portable build. Screenshots use synthetic finances.}

Understand the signal. Design around the constraints. Make the result inspectable.

I like working across the hardware–software boundary: keeping the data path clear, testing failure cases, and documenting what a result does—and does not—establish. My current focus is stronger FYP verification and repeatable evidence workflows across my projects.

Open to graduate and early-career opportunities in embedded software, Edge AI, test automation and engineering systems.

LinkedIn ↗  ·  All repositories ↗

_{Sense precisely. Model clearly. Decide locally. Build reliably.}

_{Public contributions · Refreshed weekly · Each node represents a day of building.}