Software / ML engineer building reliable AI systems, automation, and products that survive contact with the real world.
I work mostly in Python and TypeScript, across backend systems, AI applications, integrations, evaluations, and deployment.
Website · Hugging Face · Email
Malaysia-first autonomous service desk for residential aircon operators.
Telegram text and voice → booking → evaluation → human-approved SOP improvement.
TypeScript Supabase OpenAI Google Calendar Docker
Multi-tenant financial document intelligence platform with PDF ingestion, multimodal extraction, RAG, secure data isolation, APIs, and evaluation.
Python FastAPI PostgreSQL pgvector React TypeScript
WhatsApp-first accounts-receivable workflow with invoice status, payment events, reminders, AI decisioning, and a human review queue.
AWS Amazon Bedrock FastAPI Next.js Stripe Twilio
Asynchronous GPU-backed OCR service for PDF and image processing, with task queues, Redis, Docker, and structured extraction.
Python FastAPI Celery Redis PaddleOCR CUDA
The interesting part of AI is not the demo. It is everything required to make the system dependable:
- evaluations and failure analysis
- observability, health checks, and operational tooling
- idempotency, retries, and durable workflows
- authentication and data isolation
- human-in-the-loop controls
- deployment, rollback, and recovery paths
- Stackifier — billing support and management
- Context Hunt — browser tab context scraper
- stackifier on PyPI — lightweight data collection library for WhatsApp AI agents
- BSc Data Science
- Diploma in Software Engineering
- Borneo Hackathon — Idea Category
- FutureHack — 2nd Place
- Solana Flipflop Hackathon