Hồ Gia Kiện

Manga Reading Website (php,js,Bootstrap CSS,MySQL,XAMPP)

Web-based manga reading platform

Developed a comprehensive manga platform featuring an efficient database architecture. I gained hands-on experience with JS promises and async/await while utilizing PHP PDO for secure database interactions. Dealing with the complexity of this project firsthand highlighted the importance of modular organization and reusable code—principles I now prioritize in every build.

Vietnamese Voice Recognition

Training a Whisper Large model for Vietnamese speech recognition

Fine-tuned OpenAI’s Whisper-Large-V3 on a diverse corpus of Vietnamese speech to improve Word Error Rate (WER) across regional dialects. I focused on optimizing the training pipeline for low-resource environments and implemented advanced audio preprocessing to handle background noise, resulting in a more robust and context-aware transcription engine."

Seafood Shop Windows App (C#)

.NET desktop application for shop management

This was my debut project in software development, where I first learned the importance of planning ahead and architecting systems to be future-proof. I focused on making the codebase easily expandable by implementing core principles like Dependency Injection and emphasizing code reusability from the very beginning

Sanskrit Lookup App - In Construction (Electron, React)

Declension and verb conjugation lookup tool

An ambitious desktop application that bridges complex Sanskrit morphology with modern UI. This project has been a deep dive into Finite State Transducers (FST) and computational linguistics, requiring extensive research into domain-specific grammar rules. Building this with Electron and React has challenged me to maintain high performance while handling intricate logic for real-time declension and conjugation lookups.

Yu-Gi-Oh Website (WIP) (React, TypeScript, C# .NET)

Yu-Gi-Oh Website for card viewing, meta inspection and deck construction

Heavily inspired by the website yugiohmeta.com, focusing on high-utility tooling, this project leverages a React/TypeScript frontend and a C# .NET backend to manage complex card interactions. I implemented an image-recognition feature using CV principles to parse game UI elements, paired with a robust API-driven architecture to keep the platform synchronized with the ever-evolving competitive landscape.

Analyzing student performance data (Pandas)

Developed a predictive student performance model using Multiple Linear Regression (OLS) that achieved an R-squared of 0.717, identifying academic momentum and financial stability as the primary drivers of success. I utilized Backward Elimination and Variance Inflation Factor (VIF) analysis to refine features and minimize multicollinearity, while applying Chi-square and Kruskal-Wallis tests to validate significant performance gaps across demographics. The analysis provided actionable insights by demonstrating that while gender and tuition status significantly impact outcomes, macro-economic factors like GDP have negligible correlation with individual student achievement.

Analyzing NYC Taxi trip data (Pandas, PowerBI, knime)

Analyzing NYC Taxi trip data with Pandas, PowerBI, and knime

This project involves analyzing large-scale taxi trip data from New York City using Python libraries like Pandas for data manipulation and PowerBI for visualization. The goal is to uncover patterns and insights about taxi usage, traffic flow, and passenger behavior. I also used knime to build a data pipeline that automates the cleaning and preprocessing of the dataset, allowing for more efficient analysis and visualization. This project has helped me develop strong data analysis skills and the ability to derive actionable insights from complex datasets.