afes.

Kemaman, Terengganu, Malaysia

Rahimi Rusyuddin.

Software Engineer/Full-stack Developer

I build security and data systems, and I publish research on the models behind them. At U Mobile I designed CYFER ONE, an internal platform that discovers servers across the network, consolidates them into a single asset view, and drives segmentation decisions. Alongside that I am first author on deep-learning and NLP work in IEEE Access and IJ-AI. The two halves feed each other: research keeps the engineering honest about what actually works, and shipping keeps the research honest about what actually runs.

Portrait of Rahimi Rusyuddin Mohd Nazri

Systems that run in production

Security and full-stack engineering: platforms that had to work against real infrastructure, on a deadline, and survive being handed to someone else.

Models, datasets, and papers

First-author work in computer vision and NLP, published across IEEE Access, IJ-AI and conference venues. Lightweight architectures, low-resource languages, and the datasets they needed.

A comparative study of lightweight CNN architectures for recognising 11 classes of East Coast Malaysian kuih, settling on a MobileNetV2 backbone that holds 98.17% accuracy at 58 ms inference. The classifier is paired with a dual-reading interface: a KKM nutritional lookup for verified calorie data, and Gemini 2.5 Flash for cultural context the lookup table cannot carry.

98.17% accuracy58 ms inference1,227-image dataset11 classes

Link pending publication

A lightweight CNN prototype reaching 98.17% accuracy across 11 East Coast Malaysian kuih classes, integrated with Google Gemini 2.5 Flash for calorie estimation and cultural knowledge generation.

98.17% accuracy11 classesGemini 2.5 Flash

Link pending publication

Thematic classification of Malay pantun. A regex-based preprocessing pipeline segments pembayang from maksud before case folding, tokenisation and Porter stemming; an SVM baseline on TF-IDF features is then benchmarked against a custom PyTorch TextCNN. The winning model ships behind a Flask API that lazy-loads weights to keep inference under one second.

Sub-1s inferenceSVM vs TextCNNPembayang/maksud segmentation

Link pending publication

A ResNet50 framework covering 11 varieties of East Coast kuih at 95.41% accuracy, trained on a purpose-built hybrid dataset of 1,227 images and paired with a KKM-based calorie lookup.

95.41% accuracy1,227-image datasetKKM calorie lookup

Link pending publication

The first paper in the series: a lightweight MobileNet-V2 classifier for Malaysian traditional kuih reaching 95.8% accuracy, establishing the dataset and evaluation protocol the later work builds on.

95.8% accuracy38 ms inference

Link pending publication

Where the work happened

Mar 2026 – Jul 2026

Industrial Trainee

U Mobile Sdn. Bhd.

Information Services / Information Security

  • Designed and built CYFER ONE, an internal security platform for automated server discovery, asset consolidation and network segmentation: concept through production deployment.
  • Ran scheduled internal Nmap sweeps to map active hosts, open ports and exposed services, then documented and reported findings to the security team.
  • Triaged vulnerabilities out of scan output, prioritised by severity, and tracked them through to closure.
  • Worked with Information Security to fold authentication/RBAC, compliance reporting and AI-assisted querying into the platform.
  • Supported DevOps work: CI/CD pipeline maintenance, Docker workflows, shell and Python automation, and cloud operations.
  • Presented the completed system to the CIO.
Next.jsTypeScriptPostgreSQLPrismaPythonNmapDockerMicrosoft Graph

2021 – Present

Ongoing

Founder & Lead Developer

AbeFiwan Experts' Studio (AFES)

Registered web development studio, SSM 202603170295

  • Deliver and maintain live full-stack applications for SME and university clients across e-commerce, sales, booking and management use cases.
  • Built the AbeFiwan e-commerce platform on PHP/MySQL with Google APIs, including a Google Sheets backing store for one product line.
  • Built Legacy Kuih and AbeFiwan Auto Sales System 2.0 on PHP with an Oracle DB backend.
  • Built the Ezakat UiTM zakat-management system in Java against Oracle DB.
  • Scope requirements directly with clients, deploy production sites, and handle ongoing maintenance.
PHPMySQLOracle DBJavaScriptJavaGoogle APIs

Sep 2023 – Mar 2024

Industrial Trainee

BizCopier Solutions Sdn. Bhd.

IT Support & Installation

  • Led IT support for subscribed clients, diagnosing printer hardware and software faults.
  • Managed on-site installation and configuration for new corporate and retail clients.
  • Produced over 200 promotional graphics a month for web and marketing campaigns.
HardwareNetworkingWindowsCanva

Oct 2026 – 2028

Master of Science, Computer Science

Shah Alam · CS750

Expected

Mar 2024 – Jul 2026

Bachelor of Computer Science (Hons.)

Kuala Terengganu · CDCS230

CGPA 3.55 · Dean's List every semester

Oct 2021 – Mar 2024

Diploma, Computer Science & Mathematics

Kuala Terengganu · CS110

CGPA 3.19 · MUET Band 3.5

The toolkit

Languages

PythonTypeScriptJavaScriptJavaC++PHPMATLABSQL

AI & Machine Learning

TensorFlowPyTorchCNN architecturesTransfer learningMalayBERT / TextCNNMatrix factorisation (ALS)Gemini APIAnthropic SDKLLM tool-calling

Web & Frontend

Next.js (App Router)ReactNode.jsTailwind CSSREST APIsThree.jsFlaskMicrosoft Graph API

Data

PostgreSQLMySQLOracle DBPrisma ORMTemporal data modelling

Security

NmapNetwork discovery & port scanningVulnerability triageNetwork segmentationSSH-based remote scanningNextAuth / RBACbcryptMFA (TOTP)Security log review

DevOps & Cloud

DockerKubernetesCI/CD pipelinesShell & Python automationCron schedulingAzureGCPLinux / RedHat

Testing & Tooling

VitestZodGitGitHubGitLabAgile / Scrum

Hardware & IoT

ArduinoESP32C++ firmware

Recognition & leadership

2026

Winner, Best Project

GodamSahur Hackathon (wiser.my)

2025

Winner, Best Project

COMMAX (Computer and Mathematical Sciences Project Exhibition)

2025

2nd Runner Up, Best Commercialization Idea

Product Idea Innovation (PRIDE) Showcase 11.0

2024 – 2026

Dean's List, every semester

Bachelor of Computer Science (Hons.), UiTM

Aug 2024 – Mar 2026

Rural Engagement & Student Community Outreach Mission (RESCOM)

President

Ran club operations end to end (events, budget, member coordination) and acted as the liaison between the club, university administration and external communities.

Oct 2025 – Mar 2026

Service-Learning Malaysia (SULAM): Hack the Future 3.0

Project Head

Directed a school outreach programme introducing students to computer science through hands-on workshops, from securing school partnerships through to post-event analysis.

Apr 2025 – Sep 2025

Product Idea Innovation (PRIDE) Showcase 11.0

Project Head

Led the RepoBin technology-blueprint prototype and business plan. Took 2nd Runner Up for Best Commercialization Idea.

Apr 2025 – Sep 2025

Webinar Series: Artificial Intelligence and Cyber Security

Project Head

Organised and ran a webinar drawing 68+ attendees: recruited industry keynote speakers and managed promotion and registration.

May 2025 – Jun 2025

The Marvels of AI Journey

Moderator

Moderated a multi-part webinar series for students across Malaysia and abroad, running live Q&A and audience engagement.

2025

Instant Food Recommender

Collaborative-filtering recommender built on matrix factorisation (ALS) with a desktop GUI.

PythonALSMySQL

2025

RepoBin

IoT smart-bin prototype built for the PRIDE showcase, an award-winning technology blueprint.

ArduinoESP32C++

2025

Brain Tumor Detection

MRI classification model with a browser-based interface for uploading and scoring scans.

MATLABHTML

2024

Sport Classification

CNN image classifier across sport categories, presented through a lightweight web interface.

PythonCNNHTML

Let’s talk

Engineering roles, research collaboration, or a project that needs building: the form reaches me directly, and I answer everything.