Skip to content
Zafran Sakowi
zafran.dev/Johor Bahru, Malaysia

I build high-stakes
digital systems.

From migrating mission-critical Quality Control infrastructure at Mattel Malaysia to shipping full-stack B2B platforms at Cipta Craft Solutions. I specialize in building products that are fast, reliable, and maintainable at scale.

90s→5s

Dashboard load time — Mattel QC system

View ProjectsGet in Touch
01 / About
The Student

Studying Software Engineering at UTM, Johor Bahru. Started with small projects at 3am — learning every framework imaginable, from Spring Boot APIs to React SPAs. Shipped it all, badly at first, then significantly better.

UTM · Johor Bahru
The Engineer

Specialized in delivering Independent B2B Solutions for SMEs. Engineered XFitness — a unified gym management platform reducing administrative overhead by 50%, and shipped full-scale corporate architectures for Anjung Meriah Sdn. Bhd.

Client Work — Production Systems

My focus is the full stack — the interface must be pixel-perfect AND the infrastructure must be resilient. During my internship at Mattel Malaysia (MMSB), I cut Quality Control dashboard load times from 90 seconds to 5 seconds by refactoring legacy SQL workflows and engineered a tamper-proof audit log preserving 100% historical transaction accuracy.

I self-host on an Intel i3-8100 Ubuntu Server 24.04 cluster, experimenting with Docker, k3s, and automated CI/CD pipelines. Outside the terminal: documenting technical builds on YouTube (launching soon) and finishing my Final Year Project — FloraScan, an AI crop diagnostic system powered by EfficientNetB0 and a Hugging Face LLM.

Languages · Frameworks · DevOps

  • TypeScript
  • Java
  • Python
  • PHP
  • Classic ASP
  • T-SQL
  • Next.js 15
  • React 19
  • Spring Boot
  • FastAPI
  • Laravel
  • Docker
  • Ubuntu Server
  • Hetzner
  • Cloudflare R2
  • JWT Auth
Download Resumezafransakowi@gmail.com
02 / Experience
  1. Jul 2025 — Present

    Independent Software Engineer (Full-Stack) · Cipta Craft Solutions

    Delivering end-to-end B2B solutions for industrial clients and SMEs. Managed the full SDLC for XFitness (https://xfitness.my) — a gym management platform reducing overhead by 50%. Engineered custom CMS architectures and corporate portals using React 19 and Next.js 15. Integrated multi-gateway payment processing (Revenue Monster, Razorpay) and high-performance real-time features.

    • Next.js 15
    • React 19
    • TypeScript
    • Supabase
    • PostgreSQL
    • Cloudflare R2
    • Docker
  2. Aug 2025 — Jan 2026

    IT Intern — Software Developer · Mattel Malaysia Sdn Bhd

    Modernized mission-critical Quality Control (QC) systems for high-volume manufacturing. Reduced dashboard load times from 90s to 5s by refactoring legacy SQL workflows. Engineered a tamper-proof audit log system ensuring 100% data integrity for QC protocols. Successfully migrated legacy records while preserving 100% historical transaction accuracy.

    • Classic ASP
    • T-SQL
    • SQL Server
    • JavaScript
    • VBScript
View Full Résumé
03 / Projects

Cipta Craft · Lead Developer

XFitness

Payment Integration

Revenue Monster · Razorpay

Real-time Chat

Trainer ↔ Member messaging

QR Access Control

Keycard-free gym entry

Row-Level Security

Supabase RLS enforcement

Comprehensive gym management ecosystem integrating memberships, payments, QR-based access control, and trainer booking into a unified web + mobile platform. Consolidated multiple operational systems, reducing administrative overhead by 50%.

50%reduction in admin overhead
  • Next.js 15
  • React Native (Expo)
  • Supabase
  • PostgreSQL
  • Cloudflare R2
  • Razorpay

Cipta Craft · Full-Stack Dev

Anjung Meriah CMS

admin dashboard
Projects
12CRUD ✓
Services
8CRUD ✓
Properties
24CRUD ✓
Promotions
5CRUD ✓
Media Files
147Upload ✓
Audit Logs
∞Append-only

Full-stack corporate platform with a public website + secure admin CMS. JWT auth, rate limiting, ISR, and a database-driven content model — no developer intervention required for daily operations.

  • Next.js
  • TypeScript
  • PostgreSQL
  • JWT
  • shadcn/ui

Final Year Project · ML Dev

TensorFlowEfficientNetB0

FloraScan

AI-powered papaya crop diagnostic tool. Two-stage pipeline: EfficientNetB0 image classification for disease detection, followed by a Hugging Face LLM generating human-readable treatment recommendations for farmers.

  • 2-stage ML pipeline
  • Python
  • FastAPI
  • TensorFlow
  • Docker
  • Hugging Face

Pipeline

01
Image input
Real-world variability
02
EfficientNetB0
Disease classification
03
HF LLM layer
Treatment generation
04
Mobile API
FastAPI response
Containerized · Docker

Academic Project · Full-Stack Dev

TVPSS Management Information System

Centralized system for managing school TV station operations across Johor. Role-based access for administrators, teachers, and student crews. Covers scheduling, equipment inventory tracking, and automated student onboarding workflows.

  • Java
  • Spring Boot
  • Hibernate
  • MySQL
  • REST APIs
35%
reporting efficiency improvement
Layered architecture · ORM · RESTful APIs
View All Projects on GitHub
04 / The Lab

Technical Breakdowns

Deep-dive videos covering full-stack development, homelab infrastructure, and machine learning — no fluff, only code.

Notify Me
Infrastructure

Building a Self-Hosted k3s Cluster on Bare Metal

Coming Soon

Channel launching 2025

k3sUbuntuHelm~32 min

Building a Self-Hosted k3s Cluster on Bare Metal

Full walkthrough: from bare Ubuntu Server 24.04 to a production-grade k3s cluster with Helm, Traefik ingress, and cert-manager.

Full Stack

Next.js 15 & Supabase: Full-Stack App in One Weekend

Coming Soon

Channel launching 2025

Next.js 15SupabaseTypeScript~48 min

Next.js 15 & Supabase: Full-Stack App in One Weekend

Build a complete SaaS product — auth, database, storage, and edge functions — using the latest Next.js App Router patterns.

Machine Learning

FloraScan: Training MobileNetV2 for Plant Disease Detection

Coming Soon

Channel launching 2025

PythonTensorFlowFastAPI~25 min

FloraScan: Training MobileNetV2 for Plant Disease Detection

End-to-end ML: dataset prep, transfer learning with MobileNetV2, FastAPI serving, and React Native client integration.

YouTube Channel

Subscribe to get notified when the first breakdown drops.

Infrastructure deep-dives. No ads. No fluff.

Subscribe