Full-Stack Software Engineer building enterprise applications, self-hosted infrastructure, and cross-platform mobile experiences. I design backends for long-lived production systems, not demos — the software I ship serves real users and is maintained over time.

I use modern AI tooling to accelerate code reviews, refactoring, documentation, and architecture exploration — while remaining responsible for every implementation and technical decision. I care about documentation, and I usually start from the business problem itself: mapping real-world processes — quotations, approval flows, permissions — into maintainable systems. I consistently prefer production-ready, usable software over throwaway prototypes.

Outside of work, I enjoy building homelab infrastructure, running community gaming servers, experimenting with Apple ecosystems through Hackintosh, and finding ways to make software more reliable and enjoyable to use.

Interested in: secure-by-default backend systems • RBAC & auditability • enterprise software • self-hosted infrastructure • production operations

• Quotation & approval platform — an enterprise system for building, versioning, and approving quotes, used daily by a real business.
• Phố Game Chill — a cross-play (PC + Mobile) Minecraft community server for the ~1,500-member gaming community I founded and run.
• Discord community bots — music (Lavalink), ticket/support systems with payment integration, moderation, and party games (Werewolf, lottery), built in Python, TypeScript & Java.
• Native iOS app — built with Swift / SwiftUI, including a full self-configured build-and-deploy pipeline.
• Self-hosted homelab — a multi-node Proxmox + Docker + Coolify + Cloudflare Tunnel stack running my services end to end.

Languages

Backend

Databases

Frontend

Mobile

DevOps & Infrastructure

I structure backends as a layered, modular architecture and apply Clean Architecture principles where they earn their keep: HTTP concerns stay at the route layer, business rules live in services, and data access goes through Prisma. This keeps enterprise logic — multi-stage approval workflows, RBAC / role-permission models, and the full quotation lifecycle (draft → version → review → approve/reject) — testable and independent of transport and storage.

flowchart LR
    A[Client] --> B[Express Routes]
    B --> C[Service Layer]
    C --> D[Prisma Client]
    D --> E[(PostgreSQL)]

    B -. Zod validation .-> B
    C -. RBAC / Approval workflow .-> C
    C -. Background jobs / SSE .-> C

A rarer, high-value specialty I've built out is dynamic Excel automation via raw OOXML — stitching multiple spreadsheet templates together at the XML level to produce complex, brand-accurate exports that off-the-shelf libraries can't express. Combined with server-side PDF generation, this powers the document engine at the heart of the quotation platform.

Every project below is public — click through and read the code.

Also public: Hackintosh-Y9000P — OpenCore 1.0.7 EFI for a Lenovo Legion Y9000P (i9-14900HX): ACPI/DSDT patching, device-property injection and hardware bring-up notes.

Private repositories (available on request): a Discord community bot stack (TypeScript + Lavalink + Redis + Postgres, with a five-layer self-healing watchdog for voice-node recovery), a Flutter + Fastify ambient-presence app, an English-learning app, and an internal document-conversion tool for accountants with remote versioning.

• Native iOS app (Swift/SwiftUI)
• Dockerize all self-hosted services
• CI/CD pipelines (GitHub Actions) — lint/typecheck/test/coverage/audit gates on quanly-baogia; hnag runs five workflows including SSH deploy and a Fastlane iOS build on a remote macOS host
• Kubernetes — Helm chart with ingress + NetworkPolicy for the quotation platform; Argo Rollouts canary manifests with Prometheus auto-rollback for HNAG
• Migrate a production service off docker-compose onto a real cluster
• Microservices architecture
• Deeper AI integration