Application platforms
Compose and Kubernetes stacks with pinned versions, health checks, backups, and a documented restore path.
phapawith@infra:~$ whoami # Infrastructure & Platform Engineer
Phapawith.
I build the systems everything else runs on.
From bare metal to the services on top. I build and run infrastructure that is reproducible, documented, and ready to be restored.
[ OK ] bare-metal · platforms · data · aibuilt to run. documented to last.
About
I work across the whole stack: the hardware underneath, the network between, and the services people use every day.
I take care of provisioning, storage, containers, databases, and data pipelines — along with TLS, identity, backups, and monitoring.
I care about what happens after a service starts. Can it be understood, changed safely, and restored at 3 a.m. by someone who isn't me? That's the standard I work toward.
Skills
A practical toolkit. From the host to the application.
Infrastructure & virtualization
Containers & delivery
Networking
Data & databases
Platform services
Automation & AI tooling
What I build
Useful services. With the operations thought through.
Access & edge
Single sign-on, private overlay networks, automated certificates, and reverse proxies in front of services.
Data platforms
Scheduled collectors feeding warehouses, dbt models shaping the data, and dashboards making it useful.
Upgrade pipelines
Version bumps that are proposed, tested, and gated before deployment. Small, reviewed changes with a way back.
Observability
Metrics, logs, and inventory in one place, with live network and hardware topology discovery.
AI infrastructure
GPU hosts, model serving, LLM gateways with quotas and cost tracking, and retrieval pipelines on private data.
How I work
Understand before changing.
Read, propose, confirm, then act. Production changes should never be a surprise.
One change at a time.
Separate the migration from the upgrade so each step can be rolled back.
Make it reproducible.
Pin every version. No floating tags or guesswork about what is running.
Automate the second occurrence.
The first time is work. The second time is a script.
Leave a useful record.
Document as I go. Runbooks and journal entries should help the next person.
Prove the restore.
A backup is only useful if it can be restored. Test the path back.
Contact
Let's talk systems.
Infrastructure, platforms, data, or a system that's misbehaving in an interesting way. Those are conversations I enjoy.