2+
YRS EXP
5
LANGUAGES
25+
PROJECTS
20
CERTS
[ WASM OS NODE // V86 ]

BOOT
LINUX

Browser x86 emulator booting a Linux ISO through WebAssembly. All assets served locally — no external dependencies.

PROFILE
LINUX
PAYLOAD
~5.4 MB
[ EMULATED TERMINAL // X86_COMPAT ]
STANDBY
86
> wasm runtime idle
> disk image detached
> press boot to mount linux
[ INTERACTIVE MINI LAB ]

CLICKABLE
FULL-STACK PROOF

Three tiny interfaces showing how mobile clients, web dashboards, and backend services fit together. No external API here — just a fast local simulation of product thinking from screen to server.

[ SELECT PROGRAM ]
[ ACTIVE ]
Flutter pocket flow

Tap through feed, order, and chat states like a production mobile client.

EVENTS
#0
STATUS
ready
09:00 FLUTTER
[ MOBILE OUTPUT ]
CLIENT STATE

Tap phone cards. UI state updates like a Flutter app handling feed, POS, and chat flows.

[ WEBGPU COMPUTE // 2026 ]

GPU
COMPUTE PASS

Tiny browser compute lab. It checks navigator.gpu, runs a lightweight GPU pass when available, then falls back to a CPU simulation without loading heavy files.

[ COMPUTE CONTROL ]
NAVIGATOR.GPU
CHECK
MODE
idle
WORKGROUPS
--
BUFFER
--
FRAME
--
[ PASSKEYS // WEBAUTHN FLOW ]

PASSWORDLESS
AUTH CEREMONY

Production-style passkey flow simulator: server challenge, device biometric unlock, signed assertion, backend verification. No real credential is created.

[ AUTH CONTROL ]
CEREMONY
idle
LATENCY
--
CHALLENGE
CHLG-0000
AUTHENTICATOR
PLATFORM
PASSWORD
NONE
01
CHALLENGE
02
BIOMETRIC
03
ASSERTION
04
VERIFY
[ ASSERTION OUTPUT ]
credential pending
[ 2FA CONTROL // TOTP ]

Second-factor simulator for legacy login hardening: generate one-time code, enter code, verify session.

STATUS
idle
CLOCK DRIFT
0s
METHOD
TOTP
CODE
000000
PASSWORD
PLUS 1
2F-01
ISSUE CODE
2F-02
ENTER CODE
2F-03
VERIFY
[ 2FA OUTPUT ]
waiting for second factor
[ LOCAL AI NODE // TRANSFORMERS.JS V3 ]
[ BROWSER ML // NO BACKEND ]

SEMANTIC
ROUTER

Transformers.js turns requirements into embeddings, compares them against portfolio skill vectors, then routes the work to the best technical layer.

INPUT BUFFER
UTF-8
CHARS: 87
STATUS
idle
MODEL
MINILM
DEVICE
WASM
[ INFERENCE OUTPUT ]
awaiting route
MATCH
--
PIPELINE
FEATURE-EXTRACTION
RUNTIME
TRANSFORMERS.JS
[ TECHNICAL STACK ]
[ OPERATING LAYERS ]
MOBILE
Flutter + Dart production apps
APP STORE READY
WEB
Vue / Nuxt / Svelte interfaces
PIXEL PUSHED
API
Golang services with Echo/Fiber
LATENCY HUNTED
OPS
Docker + Jenkins + AWS S3 + GCP + Cloudflare
DEPLOY SURVIVED
LANG
Golang, Dart, TypeScript, Python, JavaScript
BACKEND
Golang, Echo, Fiber, FastAPI, Django
FRONTEND
Flutter, Vue, Nuxt, Svelte, React
DATA
PostgreSQL, MongoDB, Redis, MySQL
DEVOPS / CLOUD
Docker, Jenkins, AWS S3, GCP, Cloudflare, GitHub Actions
[ WEBMCP BEST PRACTICES // CHROME AI ]

AGENT
TOOL DESIGN

Best practices for building WebMCP tools that AI agents can use reliably. Derived from Chrome's developer documentation on tool strategy, naming, compute, reliability, and evaluation-driven development.

01
TOOL STRATEGY
One tool = one function. No overlap = no agent confusion.
Register tools when useful, unregister when not. More tools = bigger context window hit + slower completion.
02
NAMING
Distinguish execution vs initiation in tool names.
create-event fires now. start-event-creation-process redirects to form. Positive descriptions only.
03
MINIMIZE COMPUTE
Accept raw strings. Don't ask model to do math.
Declare specific param types: string, number, enum. Use natural labels not IDs.
04
RELIABILITY
Rate limits: return meaningful error or tell user to do manually.
Update UI state after function completes — agent uses UI to plan next steps. Validate strict in code, loose in schema.
05
EVALS
Use eval-driven dev: repeatable process, catches regressions.
Define: problem, baseline, ideal result, evaluation method. Don't patch narrow model quirks with narrow rules.
[ EXPOSED TOOLS ]
[ TOOL CONTRACT ]

Return current profile summary for browser agent context.

[ TOOL CALL LOG ]
{
  "id": "tool-call-001",
  "tool": "QUERY_PROFILE",
  "status": "ok",
  "input": {
    "scope": "profile",
    "format": "compact"
  },
  "output": {
    "name": "Watchakorn Buddeewong",
    "role": "Full Stack Developer",
    "stack": "Flutter + Vue/Nuxt/Svelte + Golang"
  },
  "ts": "READY"
}
AGENT: BROWSER
RUNTIME: LOCAL MOCK
1 CALLS
[ CURRENT ROLE // 2026 ]

FAKDUAI LOGISTICS
& DIGITAL PLATFORM

Lead developer and architect for Pinto social media super app and mobile POS system. Designed full project architecture, mentored juniors, shipped to both App Store and Google Play. Built with Go (Echo), Flutter BLoC, Vue/Nuxt, PostgreSQL, MongoDB, Redis.

[ PREVIOUS // JUN-DEC 2024 ]

IBOTNOI

Backend developer on large-scale educational platform with millions of users. Migrated Python services to Go (Fiber) reducing crashes by 50%. Engineered bulk onboarding processing 5,000 registrations/day via AWS SQS.