AI & GenAI Systems
- AI Applications
- LLM-Powered Systems
- RAG & Retrieval
- Agentic Workflows
- AI Integrations
- Evaluation & Reliability
I build production-oriented AI applications, LLM-powered systems, and scalable backend platforms, combining 6+ years of full-stack engineering experience with modern GenAI architecture.

About
Full-Stack & AI Engineer with 6+ years of professional experience, based in Malaysia.
My work spans backend architecture, API design, and frontend systems across logistics, oil & gas, e-commerce, and enterprise software, primarily with Node.js, Python, Laravel, React, Vue, PostgreSQL, MongoDB, and MySQL.
I build production-oriented AI applications alongside the full-stack platforms that support them: LLM-powered products, RAG and retrieval systems, conversational interfaces, and agentic workflows designed for reliability and deployment.
Open to roles when the fit is right.
Tech
Technologies organized by engineering domain.
Expertise
Two core areas of engineering practice, applied across production client work and personal builds.
Career
6+ years shipping production software with ownership across backend architecture, APIs, full-stack delivery, and AI integrations where the product requires them.
2021
Present
Mar 2021 to Present
Synesis Labs (formerly AlphaSquared) · Full Time · Remote
Grew from Full-Stack Developer to Senior Software Engineer by delivering production-ready features across logistics, analytics, and AI-assisted platforms.
Designed and built secure RESTful APIs with Node.js, NestJS, Laravel, and Python, focused on performance, access control, and maintainability. Improved API response times by up to 40% through query optimization and database indexing on live production systems.
Built and integrated AI-powered features using LLM APIs, conversational workflows, prompt orchestration, voice transcription, and synthesized audio responses, shipped to real users in customer-facing applications.
Built scalable, reusable component systems in React and Vue, integrated with backend APIs and real-time event workflows using WebSockets and Pusher.
Owned CI/CD pipelines with GitHub Actions and Docker, conducted code reviews, mentored junior engineers, and contributed to architecture decisions across the team.
2019
2021
Nov 2019 to Feb 2021
NetRoots Technologies LLC · Full Time · Onsite
Built and maintained enterprise business automation platforms: HRM, ERP, and CRM systems used across HR, finance, procurement, and inventory operations.
Developed backend modules from scratch and enhanced existing ones to improve operational efficiency; integrated payment gateways, cloud APIs, and notification systems.
Worked end-to-end from development and testing through deployment, and collaborated on technical documentation for implementation and QA. Strong foundation in PHP, Laravel architecture, production debugging, and systems that real businesses depend on daily.
Case studies
Representative systems from client and enterprise engagements, with live links where available.
Oil & gas characterization platform: data digitization, ingestion, visualization, and ML-driven fluids property predictions. Vue.js frontend with Python backend for data standardization and analytics.
Sports social-commerce platform combining user profiles, team chat, events, tournaments, and Stripe-powered shop with admin analytics. Laravel API with Vue.js frontend and role-based team administration.
Field service management platform for booking, technician dispatch, invoicing, and route planning. React frontend with Express API, MongoDB persistence, and Google Places integration for client reviews.
Engineering
AI and full-stack builds: RAG systems, LLM applications, SaaS platforms, realtime systems, and APIs, with source on GitHub and live demos where available.
Full-stack RAG platform for document Q&A. A Next.js 15 client handles PDF upload, file selection, and chat; the FastAPI backend handles text extraction, chunking, ChromaDB vector indexing, and grounded generation via Groq or OpenAI, with responses always tied to cited source chunks.
MeshDesk is a unified workspace for support and team communication: an AI assistant for drafting answers alongside real-time team chat, with one login and one sidebar to move between modes. The backend is a hardened Express/MongoDB API with encrypted, database-backed credential management, a pluggable multi-provider LLM layer (OpenAI, Anthropic, Groq, Gemini) with per-thread model selection and automatic failover, and a full admin panel for managing users, models, and integrations. The frontend uses a genuine design-token system with mode-aware theming, dark/light support, centralized error handling, and Pusher-powered realtime presence and typing indicators.
Voice-driven assistant: browser recording, server-side transcription, LLM response generation, and synthesized audio playback. React client with Express API orchestrating a modular voice pipeline.
A provider-agnostic LLM safety evaluation framework testing jailbreak resistance, prompt-injection vulnerability, harmful-content refusal, hallucination, and refusal consistency across OpenAI, Anthropic, and Gemini models through one unified interface. Every run is logged in structured JSONL and rolled up into CSV/JSON/Markdown reports with score-distribution charts, and the framework's own scoring and reporting logic is covered by a pytest suite.
Turborepo monorepo portfolio: Next.js apps, shared UI/config packages, MongoDB visitor analytics, and admin tooling in a dark-first marketing site.
Multi-tenant SaaS analytics and control-plane platform: a Vue 3 dashboard for revenue, pipeline, and account insights combined with tenant-scoped admin operations. Express API with PostgreSQL, JWT auth, RBAC, and tenant isolation.
A full-stack investment portfolio management application covering authentication, investments, transaction history, and analytics. The backend is a modular Express/TypeScript API organized by feature, each following a controller → service → repository pattern with Zod validation, sitting behind Helmet, CORS, rate limiting, and JWT auth with revocation, documented with Swagger UI. The React + MUI frontend uses TanStack Query and Recharts for data fetching and visualization.
Contact
Share the problem, constraints, and what done looks like. I will reply with honest feasibility and next steps.
Prefer async? Drop a message via the contact page.