Blueprints for
systems that think.
I help you see how real AI systems are designed, understand why they're built that way, and learn to build your own.

Tools I build with
- Python
- AWS
- Azure
- LLM APIs
- Vector DBs
- Agent frameworks

Build a multi-tenant SaaS on AWS
Workspaces, billing, events and an AI agent on serverless AWS, from git clone to your own module.
- 5 lessons, from git clone to your own AI agent
- Every step tested on a clean copy, with its limits stated plainly
Latest writing.
Long-form notes from building AI systems, with real numbers and diagrams.
· 6 min read
Build a multi-tenant SaaS on AWS · Lesson 1 of 5
A Multi-Tenant SaaS Starter on AWS: What's in It and Who It's For
What a multi-tenant SaaS starter on AWS includes, how it fits together, who it's for and what it can't claim yet.
Read· 6 min read
Build a multi-tenant SaaS on AWS · Lesson 2 of 5
From Clone to a Running Stack: Deploying the SaaS Template on AWS
Deploying a fresh copy of the template to an empty AWS account: the exact steps, what broke, and what the test didn't cover.
Read· 6 min read
Build a multi-tenant SaaS on AWS · Lesson 3 of 5
How Multi-Tenancy Works in a Serverless SaaS on AWS
How the template keeps customers apart in shared tables: choosing a model, the login token and tenant keys.
ReadI build the part that comes after the demo.
Retrieval pipelines, multi-agent orchestration, and the evaluation and monitoring that keep both dependable. I run these workloads on AWS and Azure and care most about the gap between a working prototype and a system people rely on.
This site is where I write down what I learn while closing that gap.
- Fullstack AI Engineer
- AWS Certified AI Practitioner
- Computer Science, Hanoi University of Industry
- Based in Hanoi, Vietnam

Have an AI system to build?
I help teams take AI from idea to production. Here's where I can help.
RAG over your data
Search and Q&A over your documents, with answers that cite their sources and retrieval quality measured, not assumed.
Agents that use your tools
Multi-agent workflows wired into your APIs, with guardrails on what each agent may do and fallbacks when it fails.
Demo to production
Taking a working prototype live on AWS or Azure: evals, tracing, cost and latency control, and safe deployment.