Blog
Thoughts on building AI products, data engineering patterns, and lessons from shipping software at startups and enterprises.
I Got Locked Out of My Own Server During Chinese New Year. Then I Automated Everything.
Three lockouts during a holiday engagement taught me more about Ansible, server hardening, and disposable infrastructure than any tutorial ever could. Here is the playbook that survived.
RAG Is Not Enough: How We Built Dual-Mode Document Intelligence
Why basic RAG fails for enterprise documents and how we built a dual-mode system combining passive retrieval with agentic reasoning and VLM-powered document processing.
The 5-Layer AI Guardrails System That Catches What GPT Misses
How we built Sentinel, a progressive AI safety middleware with 5 input layers and 4 output layers that balances catch rate, latency, and cost.
Why I Left Actuarial Science for AI (And Why My Stats Background Is My Secret Weapon)
A personal career journey from actuarial science to leading AI solutions, and why statistical rigor is the most underrated skill in modern AI engineering.
How We Built 15 AI Agents That Actually Work in Production
Behind the scenes of AiMod's multi-agent DataOps platform: decision routing, bounded autonomy, human-in-the-loop patterns, and the architecture lessons from building 15 LangGraph agents.
Building a 6TB Data Warehouse From Scratch: Lessons I Wish I Knew Earlier
War stories from building a PostgreSQL data warehouse at Credit Bureau Malaysia: schema design, partitioning strategy, Airflow gotchas, and performance optimization at scale.