Skip to content

วัชรพล วีระบริรักษ์

Watcharapon Weeraborirak (วัชรพล วีระบริรักษ์) — ML & AI Engineer

ML & AI Engineer at Buzzebees. Builder of Goatie.

I turn customer data into systems that make decisions.

Watcharapon “Kane” Weeraborirak looking out over a Bangkok street at night

Bangkok, after hours. Where most of the ideas start.

By day I engineer machine learning and customer data for a loyalty platform used by brands across Thailand. After hours I build Goatie, so a small business can keep its books from a chat. Same rule for both: if it does not run in production, it is not done.

What I’m working on now.

Day job

ML & AI Engineer at Buzzebees.

I own the machine learning layer of a loyalty platform: forecasting brand health, recommending rewards, and giving marketers an AI they can question about their own members.

  • Brand Health V3: forecasts each brand’s health from membership, activity and commerce signals, then explains the score in Thai and English with an LLM.
  • Reward recommendation: ranks the rewards and campaign ideas most likely to land with each segment.
  • Databricks MCP server: a governed bridge that lets ChatGPT and Claude query loyalty analytics behind OAuth.
Read the case study

Brand health, end to end

In production

  1. 1Loyalty data in Databricks
  2. 2Feature snapshots
  3. 3Forecast and scoring
  4. 4Serving API
  5. 5Bilingual AI explanation

Databricks, MLflow, Azure AI Foundry, Azure AutoML, FastAPI, PostgreSQL, MongoDB

Shipped

A customer data platform, built from the ground up.

One customer, a dozen records. I designed and built the platform that pulls customer, purchase and point data from every source, cleans the history, resolves identities into one profile and publishes it for analytics and journeys.

  • Ingestion from files, databases and daily rolling feeds into a typed history in PostgreSQL.
  • Identity matching that stitches scattered records into a single person, with every decision traceable.
  • Publication into ClickHouse for profile, purchase and point views, and immutable audience snapshots for Journey Builder.
Read the case study

From raw records to one profile

Live

  1. 1Files, databases, daily feeds
  2. 2S3 landing
  3. 3Typed history in PostgreSQL
  4. 4Quality and normalization
  5. 5Identity matching
  6. 6Profiles in ClickHouse
  7. 7Audiences for journeys

AWS CDK, AWS Glue, Fargate, Step Functions, PostgreSQL, ClickHouse, Elasticsearch, Kafka, Temporal

After hours

Goatie. Accounting that lives in LINE.

Snap a receipt and Goatie reads it, records it and files a clean PDF into your company’s Google Drive, ready for your accountant. Built with an accountant partner for Thai SMEs.

  • Receipt OCR with LLM extraction, and a manual path for expenses that never had a receipt.
  • Quotations, invoices, receipts and payment vouchers rendered as PDFs.
  • VAT and withholding-tax reports that keep a snapshot of exactly what was filed.
Visit Goatie

One photo, one clean record

Live at heygoatie.site

  1. 1LINE or web upload
  2. 2OCR and LLM extraction
  3. 3Structured records
  4. 4PDF documents
  5. 5Google Drive and Sheets
  6. 6VAT and WHT reports

FastAPI, Next.js, LINE LIFF, n8n, PostgreSQL, pdfme, Google Drive API

Inside the workspace.

More than 190 repositories, grouped the way I think about them. Pick a folder to see how each system flows.

~/PycharmProjects/customer-data

README.md

Customer data and journeys

Ingestion, identity resolution and publication for the CDP, plus Journey Builder: graph, event and schedule triggers that act on committed audience snapshots.

How it flows

  1. Sources
  2. Typed history
  3. Identity
  4. ClickHouse
  5. Journeys

Built with

AWS CDK, Glue, Fargate, PostgreSQL, ClickHouse, Kafka, Temporal, React

Tools I reach for

Languages
Python
TypeScript
SQL
Data
Databricks
PostgreSQL
ClickHouse
Kafka
Temporal
ML and AI
MLflow
Azure AI Foundry
AutoML
LLMs
MCP
Cloud
AWS CDK
Glue
Fargate
Azure
Docker
Product
FastAPI
Next.js
Nuxt
React
n8n

Selected work.

Systems I designed and shipped, at work and on my own.

Portrait of Watcharapon Weeraborirak in a black suit

Hi, I’m Kane.

วัชรพล วีระบริรักษ์ Watcharapon Weeraborirak

My name is Watcharapon Weeraborirak (วัชรพล วีระบริรักษ์), and most people call me Kane. I am an ML and AI engineer based in Bangkok.

I started out building chatbots and computer vision, moved into satellite and geospatial AI at Thaicom, and now work on loyalty data at Buzzebees. The thread through all of it is the same: take messy, real data and turn it into something a business can act on.

Outside the day job I build products of my own. Goatie is the one I care about most right now.

Based in
Bangkok, Thailand
Role
ML & AI Engineer, Buzzebees
Studied
B.Eng. Computer Engineering, Southeast Asia University
Speaks
Thai, English

Off the clock

Kane at night beside a car under street lights
Late drives
Kane in Yaowarat at night under neon signs
Yaowarat at night

How I got here.

  1. Now

    Buzzebees

    ML & AI Engineer

    Brand health forecasting, reward recommendation, the customer data platform and AI access to loyalty analytics.

  2. 2023 – now

    Independent

    Product builder

    Goatie for Thai SMEs, EasyStay for hotels, Pasa Education for online learning.

  3. From Jul 2022

    Thaicom PLC

    AI Engineer

    Geospatial AI for carbon and disaster analytics, and AI4ALL, a governed enterprise GenAI platform.

  4. 2020 – 2022

    Mango Consultant

    AI Developer

    CRM chatbots on LINE and Facebook, face-recognition attendance and RPA for enterprise teams.

  5. Before

    Southeast Asia University

    B.Eng. Computer Engineering

    Presented at the National Research Office Inventor’s Day and competed in NSTDA’s IT Contest of Thailand.

Certified.

Training and recognition

  • AWS Security Engineering on AWS, BSI Training Academy2023
  • Developing on AWS, Amazon Web Services2023
  • AWS Technical Essentials, Amazon Web Services2023
  • Inventor’s Day Presenter, National Research Office2020
  • IT Contest of Thailand, 19th cohort, NSTDA2020
  • NECTEC Innovation Programme, NECTEC2019

Questions people ask.

Who is Watcharapon Weeraborirak?

Watcharapon “Kane” Weeraborirak (วัชรพล วีระบริรักษ์) is an ML & AI Engineer at Buzzebees in Bangkok, Thailand. He builds customer data platforms, machine learning for loyalty programs, and Goatie, an AI accounting assistant for Thai SMEs.

What does he work on at Buzzebees?

He owns the machine learning layer of the loyalty platform: Brand Health V3 forecasting with bilingual LLM explanations, reward recommendation, a customer data platform with identity resolution, and a Databricks MCP server that lets AI assistants query loyalty analytics securely.

What is Goatie?

Goatie is a LINE-first AI accounting assistant for Thai small businesses. It reads receipts, creates quotations, invoices and payment vouchers, files PDFs into the company’s Google Drive and prepares VAT and withholding-tax reports. It helps prepare documents; it does not replace an accountant.

What is the customer data platform he built?

A platform that ingests customer, purchase and point data from files, databases and daily feeds, stores typed history in PostgreSQL, matches identities into a single profile and publishes profiles to ClickHouse for analytics and Journey Builder audiences. It runs on AWS with CDK, Glue, Fargate and Step Functions.

What technologies does he use?

Python, TypeScript and SQL; Databricks, MLflow, PostgreSQL, ClickHouse, Kafka and Temporal; AWS (CDK, Glue, Fargate, Lambda, Step Functions) and Azure (AI Foundry, AutoML, Bicep); FastAPI, Next.js, Nuxt and Docker.

How can I contact Watcharapon?

Email wera.watcharapon@gmail.com, connect on LinkedIn, or book a 30-minute call at cal.com/watcharap0n.

Let’s build something real.

Data platforms, ML in production, or an AI product that needs to ship. Tell me what you are working on.

Email meBook a 30-minute call

wera.watcharapon@gmail.com

LinkedInGitHubRésumé