Hafiedz Helmi
Role: Software Engineer @ JurisTech
Study: MSc Data Science (part-time)
Location: Kuala Lumpur, MY
Stack: PHP · Python · AI
Status: open to work
Building lending systems by day. Training models by night. Software Engineer at JurisTech, pursuing a data science master’s part-time, plus web apps, ML, and microcontroller builds.
Experience, 3 roles
Software Engineer
@ JurisTechBuilding features across an enterprise loan origination system for banking clients, working on the web application itself as well as the API layer that integrates it with the mobile version. Day-to-day work spans backend PHP, query optimisation across Oracle, MariaDB, and DB2, and SOAP/REST integrations, plus client deployments, disaster recovery, and on-call production support, increasingly through agentic AI development workflows.
Software Developer
@ Kasatria TechnologiesEmbedded in the internal R&D team, building tools to streamline colleague workflows: a multi-source Google Sheets automation, automated test coverage to cut manual verification, and a web-based prototype navigator with clickable hotspots to simulate user journeys without product code.
Software Engineer Intern
@ KOLLECTResolved UI and data bugs across a Laravel debt-collections platform serving multiple clients, from client meetings through to fixes and testing, and optimised slow queries. Drove a Laravel 6→11 upgrade, refactoring raw SQL to Eloquent ORM so the client could integrate with newer systems.
My cool stuff, 7 projects
DIY smart home for my room. An ESP32 hub drives the AC through an IR blaster, Yeelights, Tuya plugs and a Xiaomi purifier over their own local protocols, and shows them in Apple Home for Siri. It keeps one outgoing MQTT link to a PHP server, so a tap reaches the device in under a second without opening the home network. The web app adds scenes with timed steps, energy and climate insights, and Apple Watch sleep data. Deployed to the cloud. The demo is a made-up house that resets every hour.
My personal-finance tracker, built from scratch with a custom PHP backend and a mobile-ready Ionic React frontend. Covers accounts, net worth, goals, and PDF/image bank-statement import via GPT-4o. Features an AI agent with dedicated finance tools for forecasting, scenario planning, and spending insights, plus consolidation of live transactions from various sources. Deployed to the cloud.
Browser extension that filters explicit content in real time. Dataset built by scraping Google and Reddit using their APIs, then resized and split for training. A fine-tuned MobileNetV2 model classifies images into 10 sensitivity levels, and users set their own threshold to control what gets blurred.
Skin-disease classification model built with PyTorch. Data collected via web scraping, then multiple training iterations with architecture experiments tracked across 9 runs, each with its own confusion matrix and feature-map visualisations.
News management system with role-based access control for editors, staff, and readers. Article publishing, category management, user authentication, and email verification.
Automated orchid-watering system on ESP32. Reads temperature, humidity, soil moisture, and tank level, then uses a fuzzy inference system (27 rules, 3 inputs) to set exactly how long to run the pump. Monitored remotely via Blynk.
Collaborative tomato-leaf disease detection model. Each contributor trained their own model variant on a shared labelled dataset, comparing approaches across different architectures for agricultural image classification.
Skills
Résumé, on request
# My résumé is not a public download. Tell me who you are and what you are hiring for, and I will send a copy over myself, usually within a few hours.
# Requests come straight to me. No list, no newsletter, no automated follow-up.
# In a hurry? Email me and it is the same result.