AI PRODUCT · WORKFLOW · QUALITY

Native Thai • Working English • Bangkok, Thailand • Remote

Taninnuth Warittarasith

Reliable AI starts with clear judgment.

AI-Assisted Product Builder • AI Workflow & Quality Specialist

I turn business problems into practical AI-assisted systems by defining the rules, exceptions, quality criteria, and human-review boundaries before accelerating implementation with AI coding agents.

View Selected Work GitHub

Selected Work

Systems shaped around real operational problems

Three case studies showing workflow design, selective AI use, and explicit quality controls.

Internal

LINE OA Cafe Kiosk Ordering Engine

State-driven LINE ordering workflow with failure handling and multi-tenant operations

WorkflowState MachineIdempotency

Problem

Order, payment-slip, queue, and pickup workflows can become inconsistent when state lives across chat messages and external integrations.

What I did

Defined a store-scoped order state machine with LIFF customer flows, Supabase-backed live mode, admin RBAC, idempotency for create/slip/webhook/action paths, failed-delivery visibility and replay, and tenant bootstrap tooling.

Prototype

SlipLedger

Rule-first accounting workflow with optional AI and human review

Human-in-the-loopRule-firstAccounting Workflow

Problem

Raw statements, receipts, categorization decisions, and accounting handoff are easy to fragment.

What I did

Designed a rule-first suggestion engine with vendor memory and optional AI, transaction review, evidence linking, DR=CR validation, ledger/trial-balance/reconciliation flows, role-based access, and export packs for accountant review.

Prototype

Lead Cap / Name Card Keeping

Privacy-conscious AI extraction from business cards into user-owned Google Sheets

AI ExtractionPrivacyOCR

Problem

Trade-show lead capture is slow when business cards must be typed manually, but centralizing raw card images and lead data creates unnecessary privacy and operational risk.

What I did

Designed LINE/LIFF identity, Google OAuth/Sheets delivery, an image gate before full OCR, duplicate checks, quota/purchase state, and an architecture that avoids persisting raw business-card images or lead/customer records in the central app database.

More Work

Combat Performance Fueling — Evidence & Rule Architecture

Evidence-to-rule traceability and professional review before automation

Case Study

AI Intake Assistant

Structured intake and missing-information detection before AI-assisted handoff

Prototype

Sales Proposal / Export Document Generator

Structured proposal and PDF-generation workflow

Prototype

What I Bring

Clear judgment from evaluation through implementation

I work where AI quality, operational workflows, and practical systems need to stay understandable and verifiable.

AI Evaluation & Quality

I evaluate generated outputs against explicit goals, constraints, and context, then document clear, actionable feedback.

Output comparison
Quality criteria
Thai-language context

Workflow & System Design

I translate business operations into states, rules, integrations, and exception paths, using AI coding agents to accelerate implementation.

Requirements and rules
Failure-aware workflows
Practical integrations

Human Review & Verification

I keep people responsible for uncertain or high-risk decisions and add validation, traceability, tests, and handoff checkpoints.

Human-in-the-loop
Traceable decisions
Tests and smoke checks

Background

Engineering foundation in design and product roles, followed by business ownership and independent AI-assisted product and quality work.

2007–2009

Toshiba Carrier · Design Engineer

2009–2012

Seagate Technology · Product Engineer

2012–2013

Hana Microelectronics · Product Engineer

Now

Business Owner · Independent AI Work

Contact

Available for remote AI quality and workflow opportunities.

For roles involving generative AI evaluation, human-in-the-loop operations, or AI-assisted product work, contact me directly.

taninnuth@gmail.com

GitHub LinkedIn