Justin Ong

Logo

View My GitHub Profile

Hello there, I’m Justin!

Data Analyst

website    website

Summary of Skills

Programming Languages:
Python • Java • SQL • C++ • HTML • CSS • JavaScript • TypeScript
Frameworks:
Pandas • NumPy • scikit-learn • Keras • Matplotlib • Spring Boot
Tools:
Power BI • Power Query • Power Automate • Excel • Visio • Tableau • MySQL
IDEs:
Jupyter Notebook • Visual Studio Code • PyCharm • IntelliJ IDEA
Soft Skills:
Project Management • Technical Writing • Collaboratiion • Transformation

—

Work Experience

Data Analyst @ The Estée Lauder Companies Inc.
[Dec 2025 - Present]

I build and enhance dashboards while maintaining their data pipelines.

What I do:

 

HR Analyst Intern @ The Estée Lauder Companies Inc.
[Apr 2025 - Nov 2025]

In this internship, I led development for multiple RPA projects, reducing thousands of manual tasks into quick and effective automation processes. I also had my first experience working with real industry data, pipelining raw data into fine-tuned dashboards for management-level teams.

What I did:

Notable Achievements:

 

CS Research Assistant @ Southern Arkansas University
[Sep 2022 - Dec 2024]

I assisted in several machine learning research projects, specifically ViT (vision transformers) object detection. Python and its libraries (Tensorflow, Keras, PyTorch, CUDA) were utilized for our projects.

What I did:

 

Peer Tutor @ Southern Arkansas University
[Aug 2022 - Dec 2024]

In the Academic Enrichment Center, I worked as an in-person supplemental tutor to peers, providing tailored support to individuals who required assistance in mastering their course material. This role trained my communication and leadership skills.

What I did:

 

App Dev Intern @ J.B. Hunt Transport
[May 2024 - Jul 2024]

Internship Report • Internship Slides • Presentation Video

In this paid, in-person internship, I participated in multiple 2-week scrum sprints alongside a full-time backend team. We focused on updating and maintaining the 360 Shipper application, where I was assigned stories on EUM calls cleanup, removing and replacing legacy code with modern APIs.

The application was based on the Java Spring Boot framework, and we used Git requests to collaborate through Azure DevOps Services. We also utilized Postman API and Dynatrace to test our services.

What I did:

—

Education

Southern Arkansas University
[Jan 2021 - Dec 2024]

Bachelor of Science in Computer Science: Cyber Security and Privacy Option, Minor in Mathematics, 4.00GPA

Honors College, International Students Association

Academic activities:

Extracurricular activities:

On-campus jobs:

—

Projects

1. Spotify Wrapped in Power BI

GitHub Repo

Skills: Data ETL • Data Visualization • DAX • 3rd Party Web API

Tools: Power BI • Power Query

Overview:

Spotify Wrapped is a popular online phenomenon that pops up every year, but it is inherently incomplete, as data is often cut by mid November to generate the summary before the year ends. In this project, I reconstructed and customized Spotify Wrapped using Spotify’s provided personal data and public Web API, producing a extensive view of the classic report that can span multiple years.

Using Power Query, tens of thousands of streaming records are ingested from raw, local JSON files and transformed into a structured semantic model. After basic cleaning, unique track and artist identifiers are grouped into controlled batches to be queried to the Spotify Web API. Through the Web API, the model can obtain additional resources like album and artist images and genre information. To prevent exceeding rate limits, the most relavant tracks and artists are targeted and grouped for batch querying, reducing ~25000 single queries to just 33 API calls.

The result is a fully reproducible, end-to-end analytics pipeline that combines raw data, API-based enrichment, and analytical modeling to recreate — and extend — Spotify Wrapped with greater accuracy, completeness, and analytical flexibility.

Note: According to official Spotify documentation, API calls in the future will require a premium Spotify account.

 

2. Hobbit Face SVM Classifier

If most of us valued food and cheer and song above hoarded gold it would be a merrier world. — J.R.R. Tolkien

GitHub Repo

Skills: Python • NumPy • Pandas • Matplotlib • OpenCV • PyWavelets • scikit-learn • Keras • HTML • CSS • JavaScript

Tools: Jupyter Notebook • VS Code • PyCharm • Flask

Overview:

This project explores whether a machine learning model can distinguish between actors who portrayed Hobbits in The Lord of the Rings. Because the actors share similar visual traits, the task serves as a fine-grained face classification challenge.

I scraped images online of Elijah Wood, Sean Astin, Billy Boyd, Dominic Monaghan, and Martin Freeman that were ingested into a Jupyter Notebook setup. Faces were detected and cropped using OpenCV Haar cascades, and low-quality images were filtered out. The initial model used a traditional pipeline with wavelet feature extraction and an SVM classifier, tuned using GridSearchCV. The trained model was deployed via a Flask backend and a simple web interface that supports image uploads, multiple face detection, and confidence-based predictions.

To improve performance, the SVM was replaced with a Convolutional Neural Network trained directly on raw face images. After 20–30 epochs, the CNN achieved a ~20% improvement in accuracy over the original approach.

The final system demonstrates the effectiveness of CNNs for distinguishing visually similar faces in a real-world classification setting.

 

3. Real Estate Price Prediction

GitHub Repo

Skills: Python • NumPy • Pandas • Matplotlib • scikit-learn • HTML • CSS • JavaScript

Tools: Jupyter Notebook • VS Code • PyCharm • Flask • Postman

Overview:

In this regression project, I used a U.S. real estate dataset (2.2M+ entries) on Kaggle that was extracted from Realtor.com to create a prediction model that estimates the price of a property based on house area (square feet), number of bedrooms and bathrooms, and state.

I started by preprocessing the dataset and used it to build a model with scikit-learn using linear regression. The model was then exported as a Pickle file. Next, I created a Python Flask server to run the model and receive GET and POST requests, which I tested using Postman. Lastly, I made a webpage using HTML, CSS, and JavaScript with a user-friendly UI, where the user can enter their desired inputs to get a predicted price.

The model building section covers a majority of data science concepts like data cleaning, outlier removal, feature engineering, dimensionality reduction, one hot encoding, and K-Fold cross-validation.

 

4. KAMI (Kitchen Assistant and Meal Innovator) - AI Recipe Generator

Project Report • Project Poster

Skills: Project Management • Python • SQL • Data Modeling • AI Prompt Modeling • HTML • CSS • JavaScript

Tools: Django • XAMPP MySQL • GPT-4 • DALLE-3

Overview:

This senior capstone project was carried out across two semesters (roughly 9 months) with two of my buddies. We met twice a week physically to discuss our individual and collaborative progress with our senior project advisor. Other forms of communication and collaboration was done through Discord, Google Docs, and Google Slides.

The main focus of the project was to make an AI-based recipe generator that functions by feeding it with available ingredients and establishing limits or constraints to the dish. We integrated this into a website that ‘invents’ new recipes by considering the user’s dietary needs and choice of cuisine.

The recipe’s ingredients, instructions, and AI-generated image are displayed to the user. The project utilized OpenAI’s GPT-4 and DALLE-3 APIs into a Django Framework that is connected to a local database.

I fully designed the webpage using HTML, CSS, and JavaScript. I also engineered the entire database that handled the many-to-many relationships between users and their available ingredients, preferences, and generated recipes. The database was hosted on a local XAMPP MySQL database. Unfortunately, we took the demo website offline due to its costs.

—

Skills

Data Visualization

The following is a simple interactive data visualization of the wages of computer science jobs between 2020-2023. I created this while learning and experimenting with Tableau.

Tableau Dashboard

Here’s my simple take on the classic Tableau Sample Superstore dataset:

Tableau Dashboard

As a research assistant, I had to explore different ways of visualizing data. I utilized Python libraries and Jupyter Notebook during our machine learning projects.

Jupyter Notebook Visualization

 

Technical Writing

Sample Proposal • Sample Memorandum • Sample Recommendation Letter

I believe that technical writing skills are vital in the Computer Science field, to report documents professionally. I have written some technical documents outside of my CS projects to develop these skills.

During my Technical Writing class, I wrote a sample proposal where I sent out digital surveys on a potential product and fully analyzed and researched that topic with both primary and secondary sources. Then I designed the proposal in an aesthetic yet informative document.

Proposal Sample Page

I also had the opportunity to write a sample memorandum to my professor, analyzing the technical effectiveness of a particular product I was familar with.

Memo Sample Page

—

Certifications

Arkansas Summer Research Institute 2023

Certificate of Completion

A virtual two-week professional development event hosted by Arkansas NSF EPSCoR, where partipants learn technical and professional skills in the data science field.

—

Other things about me

In my free time, I like do digital art and animations. Creative projects have always been my interest.

I had the opportunity to design the logo/mascot of my Senior capstone project, where I even created a .gif animation consisting of 16 hand-drawn frames.

Animation frames: