From software to AI
Over more than ten years, mostly in banking, I moved from software engineering to machine learning, then data platforms, and now AI. Each stage left me with something I still use: systems must run reliably, data must be correct and owned, and models must be evaluated and monitored. AI agents need all three.
Today I am a Senior Data & AI Architect, responsible for the architecture of a data and AI platform on Databricks and AWS: MLOps, retrieval over internal documents, and a framework for AI agents and MCP, within the bank-wide enterprise architecture.
How I work
- Start from the need. Before drawing any diagram, I ask what the solution is for, who will use it, and who owns the result when it is wrong.
- Plan the hard parts early. Data quality, access control, safety, cost, and recovery are cheap to design on day one and expensive to add after launch. They go into the first sketch, not the final review.
- Build shared capabilities with clear boundaries. A prompt, a skill, a knowledge base, or a data product is worth more when other teams can reuse it. That only works when each one has a clear interface, a clear scope, and an owner.
- Test with real data. Demos run on tidy samples. I evaluate with real documents, acceptance criteria agreed in advance, and the people who will actually use the result.
Teaching and sharing
At work, I mentor and teach in internal training for data and AI teams. Outside work, I speak at technical talks and community events, most recently TechMaker Coffee Talk #2 in September 2026, on building reusable AI capabilities for enterprise data teams. A good question from the audience usually teaches me something too.
Education and certifications
Master’s degree in Computer Science, focused on data science and big data, and Engineer’s degree in Computer Science (Honors Program), both from Ho Chi Minh City University of Technology.
I also hold a number of international certifications across data, machine learning, and cloud, such as Databricks, AWS, and Red Hat.
About this site
HarryArchitect is my notebook: what I learn from designing systems, working with teams, and talking with the community. It is a personal space, not an official voice for my employer.
If you work on data, AI, or growing technical teams, I would be glad to talk. Find me on LinkedIn.