Case file — open · Software Engineer at RAASPAL

SWANHTET AUNG

Forward Deployed Engineer / Software EngineerEmbeds with the client and ships what they actually need — live sites, AI-drafted proposals, and the Spring Boot platform a robot fleet reports to every night.

Open case files View CV

File 01 — Subject profile

The Subject

Exhibit A — Swan Htet Aungverified

Computer Engineering graduate, now a Software Engineer at RAASPAL working the forward-deployed way: sit with the client, find the real problem, ship the software that solves it. So far that has meant two live robotics sites, a facility-monitoring demo for a hospital group, and the internal platform the company runs on — a Spring Boot backend with Claude-drafted proposals and telemetry from around 150 robots, behind three Next.js apps. Backend at heart, but goes wherever the job is.

Current post: Software Engineer / Product Engineer at RAASPAL (hover to declassify)

Field experience

1+ yrs

full-stack development

Training

B.Eng.

Computer Engineering

Preferred theatre

Backend

systems & APIs

File 02 — Known capabilities

The Arsenal

Cloud

07 items

The ground the city stands on. Machines nobody sees, running all night.

AWS LightsailIntermediate
AWS S3Intermediate
AWS EC2Intermediate
AWS RDSBasic
AWS IAMBasic
Vercel · NetlifyExperienced
SupabaseIntermediate

DevOps

05 items

Getting it into the field — and keeping it there.

DockerIntermediate
GitHub Actions CI/CDIntermediate
Linux serversIntermediate
Nginx · Let's EncryptIntermediate
DNS & custom domainsIntermediate

Engineering

06 items

Written alongside AI agents. Checked by someone who knows where it breaks.

Java · Spring BootExperienced
REST API designExperienced
Spring Security · JWTIntermediate
PostgreSQL · FlywayIntermediate
Next.js · ReactExperienced
TypeScriptIntermediate

AI & Integrations

07 items

The informants. Other people's systems, made to talk.

Claude Code · AI agentsExperienced
Claude APIIntermediate
AI document extractionIntermediate
Robot telemetry APIsIntermediate
Payload CMSIntermediate
Payments · OmiseBasic
Python · scikit-learnBasic

File 03 — Field operations — live

Case Files

02 in production · 01 live demo

airoboticsth.com

Live

CF-101B2B marketing site · Production

AI Robotics

AI ROBOTICS CO., LTD. — Thailand

Bilingual B2B site for six autonomous floor-cleaning robots. Every machine has its own page, backed by a comparison table, customer references, an FAQ and a quote form — quote-only, no prices anywhere.

What I built — the design system, the robot data model, the translation pipeline and the deployment.

Thai-first routing in a Next 16 proxy; the header switch persists for a year in a cookie.

Translation pipeline with a key guard that fails the dictionary build if any Thai string is missing.

Quote form over Resend, with an offline fallback so the form is testable without keys.

Machines: 06
Languages: TH / EN
Prices shown: None
Host: Vercel

Next.js 16
React 19
TypeScript
Tailwind v4
Motion
Resend
Vercel

Open live site

findmyrobo.com

Live

CF-102E-commerce storefront · Production

FindMyRobo

RAAS PAL Co., Ltd. — Thailand

Thailand's robot shop: Mammotion mowers, Gausium scrubbers, Pudu delivery robots and T-Chef cookers. A bilingual storefront, an admin panel and an embedded CMS in one deployment.

What I built — the storefront, the admin panel, the embedded CMS, and the payment and fulfilment wiring.

Payload CMS 3 embedded at /cms with drafts, version history and live preview beside the form.

Supabase Postgres, auth and storage; sessions in httpOnly cookies, never in the browser.

Omise card + PromptPay checkout and Sokochan fulfilment built, gated behind a quote-only switch.

Brands: 05+
CMS globals: 05
Mode: Quote-only
Host: Vercel

Next.js 16
Payload CMS
Supabase
next-intl
Omise
Resend
Vercel

Open live site

raaspal-bangkokhospital.netlify.app

Demo

CF-103UI/UX concept · Live demo · Simulated data

BDMS Facility Monitor

RAASPAL — concept demo for a hospital group

A presentation-grade smart-facility console for a hospital pitch, built for Bangkok Hospital (BDMS) as a demo, not an official system: chilled-water telemetry, pumps, UPS, incident handling and monthly reports, all driven by one simulation engine.

What I built — designed in Claude Design, coded in Next.js: the UX flows, the plant diagrams, the simulation engine and the report views.

One clock, one state: physical truth → edge logger → uplink → cloud store.

Incidents detected at the edge with threshold, hysteresis and debounce; resends never duplicate.

3D, isometric and schematic plant diagrams in hand-drawn SVG, with print/PDF and CSV reports.

Views: 04
Languages: EN / TH
Data: Simulated
Host: Netlify

Next.js 16
TypeScript
Recharts
SVG
Netlify

Open live site

File 04 — Restricted

Classified Ops

Restricted · internal

CF-200Platform

RAASPAL Internal Ops

One Java backend. Three Next.js fronts. The company's whole operation.

The internal platform RAASPAL runs on: a customer survey goes in and a costed robot proposal comes out, the deployed fleet is watched through vendor telemetry, customers receive performance reports by email, and the warehouse keeps its inventory honest — all on one Spring Boot backend behind three Next.js applications.

Internal system — no public deployment. Every screen below is a reconstruction running on sample data; no customer records, credentials or endpoints are exposed.

Domain modules

Flyway migrations

70+

Fleet on telemetry

~150

Frontends served

Integrations

Claude Sonnet + Opus
Gausium
AutoXing
CVTE Kava
Monday.com
SMTP
S3

Exhibit B — system map · tap or hover a node, or use the tabs below

● packets = requests in flight

SystemThree Next.js apps call one Spring Boot API; the API owns auth, the database, AI calls and every vendor integration.

Survey in. Costed proposal out.

The staff-facing console. Upload a customer survey as Excel, PDF or a photo; the AI extracts the requirements, the backend loads the approved catalogue, and two to three ranked robot options come back with fit, specs, pricing and reasoning. Pick one and the proposal writes itself against the house template, ready to print, export to PDF or download as PowerPoint.

What I built — the upload → extract → recommend → propose flow, catalogue screens, report previews, PDF/PPTX export, Thai date handling.

Upload → extract → recommend → propose, end to end, with history for every run.
Weekly and monthly fleet reports, previewable before delivery; customers open a signed link, no login.
Live CVTE C3 device status — online/offline and battery — on the dashboard.
TanStack Query owns server state; Zustand holds only the session.