Open to full-stack roles
Full-Stack Engineer · Applied AI (RAG) · Systems & IoT
I ship end-to-end — from C/C++ systems and IoT hardware, through Go / C# / Node backends and React / Next.js frontends, to production infra and applied AI. I've taken a multi-tenant SaaS from empty repo to live product, solo — with security as the throughline.
Selected work
A live, multi-tenant SaaS point-of-sale platform I took solo from empty repo to production. Security-first: JWT sessions with token-version invalidation, 6-role RBAC, Redis-backed layered rate limiting, and a forensic audit trail. Real-time kitchen display over Socket.IO, raw parameterized Postgres, self-hosted on Oracle Cloud with GitHub Actions auto-deploy.
A self-hostable, multi-channel AI support engine in Go. RAG over an FAQ base (pgvector) that answers only on a confident match and hands off to a human otherwise — with a cost-ordered provider chain, circuit breaker, and multi-tenant isolation. Telegram, Messenger, web widget, and API from one binary. Built once, reused across MiniPOS and KanTaw.
A fully-offline AI health assistant — local RAG with Ollama + chromem-go, zero cloud. Anti-hallucination guardrails: citations only from retrieved passages, deterministic first-aid kept outside the LLM, sub-15ms emergency routing. Ships as a single ~14MB self-contained binary that shares over LAN by QR.
A clinic appointment platform in C# / .NET 9 (Blazor + a separate Web API). Dual data-access — Dapper/ADO.NET for fast reads, EF Core for domain logic — with role-based access across admin, doctor, and patient.
A cross-platform C++17 CLI that finds what's holding a TCP port — and judges it: live dev server vs. stale process vs. Docker. One interface, two backends (/proc on Linux, libproc on macOS).
A Python CLI that skims 24h of logs (Docker / Nginx / files over SSH), fuzzy-groups errors, and emails a ranked daily report. Pluggable collectors, YAML config.
Experience
Singapore · Fintech & retail automation — POS, IoT & payments · 2024–2025
Toolbox
About
My range comes from shipping in very different contexts — fintech IoT hardware across 870+ retail sites, a real-time POS platform I run in production, offline-first AI for low-connectivity settings, and freelance work delivered to a client's brief.
The throughline is production judgment — threat-aware auth, forensic audit logging, and safety-critical logic kept deterministic and out of the LLM, on infrastructure I operate myself. Most of my product work ships through my org, @jasonkyawlab.
Contact
I'm open to full-stack roles and interesting collaborations. The fastest way to reach me is email.