Fullstack software engineer, backend-strong. I make distributed systems behave — usually on the first try after the third coffee.

I'm Harry — part engineer, part "why is this not automated yet" machine.

I'm AI-driven: not in the "I let the robot write my code" way (mostly), but in the "what if this whole workflow just… did itself?" way. I like building things, but I love optimizing them — if a process takes three clicks, I've already imagined the one-click version and resented the extra two.

I'm equal parts creative and logical: I'll whiteboard a wild idea at 9pm and have a boring, sane architecture for it by 9:05. Open-minded enough to try your stack, pragmatic enough to ship the simple one first. I'm a fullstack engineer who's strongest in the backend, so when something breaks I can chase it from the pixel all the way down to the query plan — and usually blame the cache on the way. Basically: I make distributed systems behave, and I do it with a to-do list and a sense of humor.

Not a badge wall — what I can actually do when something breaks.

Java · Spring Boot · Microservices · REST APIs · gRPC

Spring Boot: where I inject dependencies instead of anxiety. I design layered, testable services with clean transactional boundaries and APIs boring enough to debug at midnight.

Kafka · Redis · Async Processing · Event-Driven Architecture

Kafka lets me politely assume the network is unreliable by designing for it: idempotent consumers, dead-letter queues, and backpressure so one slow service doesn't become everyone's problem.

MySQL · PostgreSQL · OpenSearch / Elasticsearch

I read query plans for fun — which is the most honest thing about me. Indexing, N+1 hunting, and replication topologies are where "it's slow" goes to be explained.

TypeScript · React · Node.js · Chrome Extension APIs

My frontend skills are just good enough to prove the backend works — and occasionally to ship a real UI.

AWS · Docker · Kubernetes · CI/CD

Kubernetes: I pay rent in YAML now. Containerized deploys, Helm, and pipelines I trust more than a weather forecast.

Prometheus · Grafana · Datadog · Logging · Tracing

If it isn't on a dashboard, did it even error? Metrics, traces, and structured logs so on-call is a notification, not a guessing game.

User says: "The system is slow"

        ↓
Check logs
        ↓
Check metrics
        ↓
Check database
        ↓
Check Redis
        ↓
Check Kafka
        ↓
Blame the network
        ↓
Find an N+1 query
        ↓
"Fixed."

• 🤖 AI engineering — wiring LLMs into real tools, not just chat demos
• 🛠 Developer tools — things that make the build loop shorter and the debug loop smarter
• 🔁 Automation — if I do it twice, I'd rather script it
• 🌱 Open source — small, sharp utilities I wish already existed

I'm easy to build with:

• 🤝 Collaboration — happy to pair on backend systems, open-source tools, or MVPs
• 🌱 Open source — contributors and testers always welcome
• 💡 Freelance — backend performance, scalability, and observability work

Interested? Email me at hungpt591@gmail.com or open an issue on any repo — I read everything, including the silly ones.

"Write code your future self won't hate you for." And if you make it elegant — even better.