Thien Nguyen - Software Engineer
Master Computer Science at UT Dallas. I love building things and helping people.
As a software engineer student who moved from Vietnam to the United States in 2018, I thrive on pushing boundaries and embracing challenges. With expertise in programming, web development, machine learning, computer vision, LLMs, and data analysis, I'm passionate about using technology to solve complex problems and create impactful solutions. My dedication to technology extends beyond work through my home lab, where I experiment with new technologies and configurations. Constantly seeking new opportunities to innovate, I believe that true growth happens outside our comfort zones, fueling my drive to make meaningful contributions.
IBM
UT Dallas Engineering and Computer Science
OrgXPlor
VINCEF at UT Dallas
Priority Insurance Agency
Data Analytics Bootcamp
Programming Languages
Machine Learning
Frontend
Backend
DevOps
Others
I self-host for
I run these services on my own hardware to keep my data private, avoid subscription costs, and learn how to deploy, secure, and manage real infrastructure.
Drag to rotate • Scroll to zoom • Click icons for details
Selected Work
A curated selection of projects spanning web applications, machine learning, and systems design.
RoadPatcher
A dual-sensor ML mobile app that detects potholes in real-time using accelerometer and computer vision, helping drivers and city planners identify road hazards.
MindMate
An AI-powered mental wellness chatbot that provides personalized emotional support through customizable AI companions, with comprehensive admin analytics.
Healthcare Correspondence LLM
An LLM-powered assistant that extracts and summarizes patient data from scanned medical records, reducing document review time by 50% for healthcare providers.
Brain Tumor Detection Using Deep Learning
An ensemble deep learning system that detects brain tumors from MRI scans with 95.6% accuracy using multiple CNN architectures.
Predicting Red Wine Quality from Chemical Properties using Machine Learning
A machine learning model that predicts wine quality from chemical properties, identifying key quality drivers through statistical analysis.
Building Under Pressure
4+ hackathons attended. Building incredible things in 2-3 days with passionate teams.
- H
HackUTD 2025 - Lost in the Pages
Project: TOYODARichardson, Texas
-
Built AI-powered car shopping assistant that automatically contacts 100+ local dealerships using NVIDIA Nemotron LLM for insight extraction and ElevenLabs voice synthesis; aggregates real-time out-the-door pricing, availability, and special offers into a unified comparison dashboard eliminating hours of manual dealer outreach
-
Implemented agentic AI workflow with 3 specialized Python agents—Confirmation Agent (appointment scheduling), Bargaining Agent (price negotiation), and Insight Extraction Agent (deal analysis)—orchestrated through Google Gemini for natural multi-turn dealer conversations with context retention across calls
-
Developed full-stack application with React frontend, Golang backend API, FastAPI AI microservices, and Supabase PostgreSQL database; integrated Auth0 OAuth 2.0 authentication, real-time WebSocket updates for call status, and Toyota-inspired UI/UX with search session management and dealer response tracking
-
- H
HackUTA 7 - Enter the Hacker-Verse
Project: CredifyArlington, Texas
Key Achievements:
-
Developed AI-powered credit card companion that identifies optimal card selection for maximum rewards at each merchant location using Google Gemini AI
-
Integrated Plaid API and Google Places for smart location detection, spending analysis, and personalized savings insights with budget monitoring
-
Built real-time points maximization tools comparing transfer rates to travel partners with React, TypeScript, Mapbox GL, and Supabase backend
-
- H
HackUTD 2024 - Ripple Effect
Project: OptiPropertyRichardson, Texas
Key Achievements:
-
Built AI-powered property management platform using SambaNova Cloud API to optimize energy and water consumption through real-time analytics and actionable recommendations
-
Developed predictive algorithms for consumption forecasting, enabling property managers to reduce operational expenses and environmental impact
-
Integrated real-time monitoring dashboard with Next.js, TypeScript, and Material-UI delivering data-driven insights for cost reduction
-
- H
HackUTD 2023 - Golden Hour
Project: MoodTunesRichardson, Texas
Key Achievements:
-
Developed real-time facial emotion detection system using TensorFlow and Keras CNN model trained on Kaggle dataset to classify driver emotions
-
Built intelligent music recommendation engine that automatically generates playlists based on detected emotional state, eliminating manual curation
-
Created web interface with Next.js integrating backend video analysis with frontend emotion visualization for seamless real-time interaction
-
Let's Connect
Interested in collaboration or just want to chat? Send me an email and I'll respond when I can.