// Hello! I am

Mook Nattakan

Data Scientist · AI/ML · Data Engineering

// write a command to continue:

View Experience Download Resume

// live pipeline metrics

Records (k)

Quality (%)

find me on:

About Me

My Journey

I grew up in Thailand with a love for building things — from websites and mobile apps to tinkering with Raspberry Pi and Arduino boards. After completing my Bachelor of Computer Engineering at Mahidol University, I spent time as a mobile app developer in Bangkok, building real-time tracking systems and backend APIs. But I kept finding myself most excited by the data side of every project — the patterns, the models, the story behind the numbers.

That curiosity brought me to Melbourne, where I am completing my Master of Data Science at Deakin University. It has been the best decision I have made. I have gone deep into machine learning, data engineering, computer vision, and AI pipelines — processing hundreds of thousands of records, building CNN models, designing OCR pipelines, and delivering dashboards that help teams make smarter decisions.

What drives me is the gap between raw data and real impact. I want to build things that do not just work technically, but actually change how people make decisions. That is why I am drawn to organisations with a meaningful mission — whether that is improving retirement outcomes, advancing research, or making cities smarter. Right now I am excited about Generative AI, end-to-end ML pipelines, and finding a team in Melbourne where I can grow fast and contribute meaningfully.

Skills

A grouped view of my technical stack across machine learning, data engineering, cloud, analytics, and development.

AI / Machine Learning

scikit-learn
PyTorch
XGBoost
Random Forest
SHAP
feature engineering
model evaluation
cross-validation
CNNs
LSTM
LoRA

Data Engineering

dbt
BigQuery
ELT pipelines
Kestra
Terraform
Bruin

Languages

Python
SQL
JavaScript

Cloud & Tools

GCP
AWS
Git
Vercel
Trello

Data Visualisation

Looker Studio
seaborn
Folium

Computer Vision

OpenCV
Tesseract OCR
Hough Transform
object detection
image preprocessing

Web & Databases

Vue.js
Node.js
REST APIs
React Native
MySQL
PostgreSQL
Figma

Projects

Selected data and ML projects covering analytics engineering, computer vision, and applied modelling.

Retail Analytics Pipeline

Built a five-model dbt and BigQuery ELT pipeline processing 541,000 transactions and £8.18M in revenue into staging, fact, dimension, and mart layers. Delivered a Looker Studio dashboard revealing UK market dominance, year-end seasonality, and customer concentration.

#dbt
#BigQuery
#GCP
#Looker Studio

Spotify Track Popularity Prediction

Compared Linear Regression, Random Forest, and XGBoost across 113,843 Spotify tracks. Random Forest achieved the strongest performance with R² of 0.51, while SHAP identified duration, speechiness, energy, and valence as the most influential features.

#Python
#scikit-learn
#XGBoost
#SHAP
#pandas
#seaborn

Melbourne Parking Accessibility Analysis

Integrated five City of Melbourne open datasets through REST APIs and applied a 400-metre geospatial proximity model. Identified Carlton and Kensington Primary Schools as potential high-risk areas and presented the findings through an interactive Folium map.

#Python
#pandas
#Folium
#geopy
#REST API

Population-Level Sleep Pattern Analysis

Built an LSTM model combining EEG sleep-stage sequences with demographic features from 10,115 participants. Identified age as the strongest predictor of sleep efficiency and applied early stopping to control overfitting.

#Python
#MNE
#LSTM
#pandas

Facial Emotion Recognition

Fine-tuned a pretrained ResNet CNN to classify six facial emotions across 1,198 images, achieving a macro ROC-AUC of 0.797. Addressed the small, imbalanced dataset using weighted cross-entropy loss and data augmentation.

#PyTorch
#torchvision
#ResNet
#scikit-learn
#TensorBoard

Document Deskewing Pipeline

Built an OpenCV pipeline that detects and corrects tilted document images before OCR. Compared three candidate-point strategies and selected a centroid-based method that reduced runtime from approximately 0.14 to 0.05 seconds while maintaining reliable angle detection.

#Python
#OpenCV
#Tesseract OCR
#Hough Transform

GPT-2 Fine-Tuning for Biomedical Text

Fine-tuned GPT-2 on biomedical abstracts using full fine-tuning and LoRA. Reduced perplexity from 98.28 to 17.11, while LoRA trained 57 times faster and updated only 0.24% of the model parameters.

#GPT-2
#LoRA
#PyTorch
#NLTK

Medical Distribution Web Application

Built a full-stack application for managing and distributing medicines between hospitals and pharmacies. Developed a responsive Vue.js frontend, Node.js REST APIs, and a MySQL database for managing operational data.

#Vue.js
#Node.js
#REST APIs
#MySQL

Healthcare Drug Safety Analysis (openFDA)

Analyzed FDA adverse event reports for three widely-used drugs (metformin, ibuprofen, atorvastatin), pulled live via the openFDA API. Cleaned and explored over 20,000 reports to answer six questions on severity, administration route, and outcome patterns, then built and compared a zero-shot and a fine-tuned Hugging Face model to classify report severity from reaction text alone.

#Python
#pandas
#Hugging Face
#Transformers
#scikit-learn
#matplotlib
#REST API

Animation & Event Processing Application

Implemented timer-controlled animation loops for scene transitions and movement logic. Managed bitmap assets, animation scripts, and audio playback within a real-time rendering loop using structured programming principles for modular, maintainable code.

#C#
#Animation
#Event Processing

Experience

A timeline of customer-facing and software development roles spanning hospitality, mobile, and web engineering.

03/2025 – 10/2025

Data Scientist · Team Lead

Chameleon, Melbourne

Led an 11-person data science team delivering a geospatial parking study and a population-level sleep analysis involving 10,115 participants.
Designed a structured data validation process to improve the quality and consistency of the team's analytical work.
Presented project findings to stakeholders and coordinated team progress, responsibilities, and blockers.

PythonpandasLSTMGeospatial AnalysisData ValidationTrello

Data Scientist · Team Lead

Chameleon, Melbourne

Led an 11-person data science team delivering a geospatial parking study and a population-level sleep analysis involving 10,115 participants.
Designed a structured data validation process to improve the quality and consistency of the team's analytical work.
Presented project findings to stakeholders and coordinated team progress, responsibilities, and blockers.

PythonpandasLSTMGeospatial AnalysisData ValidationTrello

03/2025 – 10/2025

01/2022 – 10/2022

Mobile App Developer

Owat Pro and Quick Company Limited, Bangkok

Led requirements gathering across three teams for a React Native workforce management application.
Designed Figma wireframes and built the employee clock-in and clock-out system before project handover.
Helped develop an internal solution that was selected instead of a commercial workforce management product.

React NativeFigmaRequirements GatheringMobile DevelopmentWorkforce Management

01/2022 – 10/2022

Mobile App Developer

Owat Pro and Quick Company Limited, Bangkok

Led requirements gathering across three teams for a React Native workforce management application.
Designed Figma wireframes and built the employee clock-in and clock-out system before project handover.
Helped develop an internal solution that was selected instead of a commercial workforce management product.

React NativeFigmaRequirements GatheringMobile DevelopmentWorkforce Management

05/2019 – 08/2019

Web Developer Intern

TeleInfo Media Public Company Limited, Bangkok

Built an internal revenue and workforce reporting dashboard using Vue.js, MongoDB, and Elasticsearch.
Delivered an Excel BI analysis projecting customer acquisition targets for individual salespeople.

Vue.jsMongoDBElasticsearchExcelBusiness Intelligence

Web Developer Intern

TeleInfo Media Public Company Limited, Bangkok

Built an internal revenue and workforce reporting dashboard using Vue.js, MongoDB, and Elasticsearch.
Delivered an Excel BI analysis projecting customer acquisition targets for individual salespeople.

Vue.jsMongoDBElasticsearchExcelBusiness Intelligence

05/2019 – 08/2019

Education

Master of Data Science

Deakin University

Melbourne, Australia

03/2024 - 02/2026

Coursework and projects spanning statistical modelling, machine learning, big data technologies, and the full data science workflow from problem framing to communication of results.

Bachelor of Computer Engineering

Mahidol University

Nakhon Pathom, Thailand

08/2017 - 10/2021

Foundation in software engineering, computer architecture, programming, and applied AI. Final year projects in web development and embedded systems.

Certifications

SQL

HackerRank

AWS Cloud Practitioner Essentials

AWS

Data Engineering Zoomcamp

DataTalks.Club

Contact

Open to data science, ML engineering, and data engineering opportunities in Melbourne and remote-friendly Australian teams.

Email LinkedIn GitHub