Infrastructure automation
Using PowerShell, REST APIs, GitHub workflows, and service integrations to remove repetitive work and make routine changes safer.
I started in service desk and Tier 2 support, which gave me a useful bias: systems are only as good as the people who have to operate them at 2am, explain them during an incident, and improve them afterward. That experience still shapes how I design infrastructure.
I still spend plenty of time close to individual servers and the real production systems around them. The broader pattern in my work is building tooling around that hands-on work: automation, health checks, observability, integrations, self-service workflows, and guardrails.
I work comfortably across Windows and Linux, Azure and hybrid environments, SQL Server, PowerShell, REST APIs, Datadog, ServiceNow, ManageEngine, GitHub, and the operational glue between them.
What I focus on now
The common thread is practical infrastructure work: make the system easier to operate, easier to understand, and less dependent on tribal memory.
Practical AI and knowledge systems
Exploring AI agents like OpenClaw, Claude Code, and Codex, plus Graphify, GraphRAG, knowledge graphs, and MCP integrations that extend AI context across code, docs, runbooks, incidents, APIs, and tribal knowledge.
Observability and reliability
Designing monitoring around useful signals, service context, and the questions teams ask during incidents, while tracking when OpenTelemetry collector patterns make sense.
Internal developer platforms
Building repeatable paths around deployment, configuration, documentation, ownership, and guardrails so teams do not rediscover the same procedures every time.
How I think about engineering
My default approach is to understand the system first, then automate the parts that should not depend on memory, heroics, or someone being online.
Automate the repetitive work
If the same ticket, checklist, or escalation happens often, it is probably asking to become a tool.
Build platforms, not one-off fixes
A good fix should make the next engineer faster, not just close the immediate issue.
Make telemetry useful at 2am
OpenTelemetry gives teams a strong common language for telemetry. The architecture still has to turn that data into faster triage, clearer ownership, and better change decisions.
Prefer maintainability over cleverness
The best automation is boring enough that someone else can trust it, debug it, and extend it later.
Recent work and side projects
The projects are where the pattern is easiest to see: practical tools, real constraints, and systems that have to be understandable after the first build.
Health data warehouse
A local-first DuckDB warehouse that normalizes Garmin, lab, scale, genetics, and manual notes so personal health data can be queried across sources instead of trapped in separate apps.
Read the write-upPromo notification service
A scheduled deal-notification pipeline that computes sports-triggered offers from authoritative APIs, dedupes until expiry, audits coverage, and formats high-trust WhatsApp alerts.
Read the write-upPaddle shop playground
A small storefront-style playground for OpenTelemetry, traces, API behavior, and observability experiments without needing a production app as the test bed.
Open playgroundBai Chan prototype
A family project to digitize a Vietnamese tile game, which means translating real house rules, fuzzy memory, and player expectations into a working interactive system.
View project noteCore strengths and tools
The areas I come back to most often are Microsoft-heavy infrastructure, hybrid cloud, observability, automation, and the operational details that make those systems supportable.
Infrastructure
Windows Server, Linux, IIS, Active Directory, SQL Server, Azure, hybrid environments, identity, migrations, and production operations.
Automation
PowerShell, REST APIs, GitHub workflows, ServiceNow, ManageEngine, scheduled jobs, data normalization, and workflow glue.
Observability and context
Datadog, OpenTelemetry, logs, metrics, traces, service context, documentation systems, local LLMs, knowledge graphs, and GraphRAG.
Outside work
Outside of work, I spend a lot of time with family, stay active, and like trips that are full and hands-on. That has included an overnight cave expedition in Vietnam, motorbiking and scuba diving in Thailand, and plenty of travel that involves movement rather than just checking in somewhere.
I am learning to play guitar and like the slow, imperfect process of building a new skill. LCD Soundsystem and The Strokes are frequent go-tos. For staying active, I tend to dabble and explore: tennis, jiu jitsu, wrestling, and more recently, pickleball.