🎉 Secret Unlocked!
You found it! Viet's robotics teams qualified for 4 consecutive World Championships — that's the real Easter egg. ↑↑↓↓←→←→BA
Hi, I'm
Machine Learning Engineer and Data Scientist building intelligent systems with Python, AWS, and machine learning. Passionate about turning raw data into real decisions.
// 01. about
Hello! I'm Viet Nguyen — a Machine Learning Engineer and Data Scientist building production-ready AI systems on AWS. Pursuing an M.S. in Computer Science at Georgia Tech while completing the AWS Cloud Institute and TripleTen's AI/ML Bootcamp.
I'm currently open to Machine Learning and Data Science opportunities — feel free to reach out!
// 02. projects
A selection of things I've built.
Trained a Random Forest classifier inside Docker, containerized a FastAPI serving layer, and published a publicly pullable ML image to Docker Hub — full reviewer approval, 81.56% accuracy, 91.5% survival probability on a live Class 1 female test.
Built a stateful, context-aware chatbot for staff shift scheduling using LangChain's ChatBedrock — covering prompt templates, output parsers, message history, and CSV document injection.
Built a RAG system on Amazon Bedrock that answers questions from a company's own documents — Titan embeddings, OpenSearch Serverless, and the RetrieveAndGenerate API.
A trained Random Forest regressor wrapped in a production-ready FastAPI service — schema-validated, startup-loaded, and refined through three rounds of reviewer feedback.
Feature engineering, class imbalance strategies, and 108 hyperparameter combos to predict which bank customers are about to leave — F1 = 0.6197 on a 20% imbalanced dataset.
Random Forest classifier recommending the right mobile plan for Megaline customers — 81.8% test accuracy, beating the target by 6.8 points.
SQL + hypothesis testing to prove bad weather makes Loop-to-O'Hare taxi rides 20.6% longer — t = 6.84, p ≈ 0.
Mined 16,715 game records to build a data-backed 2017 ad strategy — hypothesis-tested platform and genre preferences across NA, EU, and Japan.
Cleaned and analyzed 4.5M order records from Instacart's 2017 Kaggle dataset — uncovered peak shopping windows, weekly reorder rhythms, and top reordered items across 206K customers.
Interactive Streamlit app exploring 51,500+ used vehicle listings — price distributions, condition vs. model year trends, and price-odometer relationships. Deployed live on Render — no setup required.
Proposed AI system improvements for autonomous VEX robots — added SpotFi localization, RGB-D 3D mapping, and P2P decentralized coordination to solve the limited field-of-view problem.
Benchmarked Spark, Hadoop, and Amazon EMR across EC2 instance types to find the optimal Big Data configuration for a real spec-management platform used in VEX Robotics.
Connected Amazon Bedrock to a frontend via Lambda and API Gateway to generate AI-powered flashcards from study notes — fully serverless, CORS-enabled, and prompt-engineered.
Tested whether highly-rated "Golden Age" TV shows also get the most IMDb votes — cleaned messy real-world data (missing values, duplicates, implicit string errors) before confirming the hypothesis.
Used SDA tools to compare Megaline's Surf and Ultimate prepaid plans across 500 clients — computed per-user monthly revenue from 5 merged tables, analyzed behavior via barplots/histograms/boxplots, and confirmed via two-sample t-test that Surf generates significantly more revenue (avg $50.33 vs $47.31, p ≈ 0).
Applied Disparate Impact and Statistical Parity Difference to 2.5M Fannie Mae mortgage records across Race and Gender — then measured whether Reweighting pre-processing bias mitigation survives classifier training. Spoiler: it doesn't.
Built a custom Terraform provider in Go with full CRUD support, a reusable AWS module (S3 + CloudFront), and deployed two static-site homes to a live cloud platform
Coming Soon
Additional ML and data science projects are being documented and will appear here shortly. Check back soon or view my GitHub for the latest work.
// 03. skills
Technologies and tools I work with regularly.
// 04. experience
TechX Robotics · Tustin, California
Managed full P&L for a $43,000+ revenue robotics education business, auditing inventory and bank records across 2 fiscal years and identifying $7,355+ in unreconciled equipment purchases. Built an LLM-assisted owner compensation model using parallel outputs from 3 AI systems (Claude, ChatGPT, Gemini) to analyze distribution scenarios — applied LLM evaluation for a real financial decision. Designed the competitive framework that led VEX Robotics teams to 1 World Championship.
Jeffrey Trail Middle School · Irvine, California
Manage an annual $15,000–$20,000 engineering and robotics program budget. Teach robotics, AI, computer science, and biotechnology to middle schoolers. Coached teams to 4 consecutive World Championships and 5 consecutive State Championships.
Association of Talent Development – Orange County (ATD-OC) · Anaheim, California
Generated a digital marketing report for ATD-OC via an AI-powered agentic workspace (AWS Quick Suite), delivering a 3-phase strategic roadmap targeting 15% membership growth and 40% engagement improvement. Built and deployed an interactive marketing analytics dashboard with Claude via Vercel to accompany the report.
Manhattan Beachwear · Cypress, California
Supported post-ransomware IT overhaul following a $10M breach — deploying MFA, AI-endpoint protection, and secure remote access across employee systems.
Countrywide Home Loans · Palm Desert, California
Originated prime and subprime mortgage loans during the height of the U.S. mortgage crisis.
// 05. certification
Amazon Web Services · Certification
// 06. education
Georgia Institute of Technology
TripleTen
Amazon Web Services
University of Southern California
// 07. contact
Whether you have a project in mind, a job opportunity, or just want to say hi — my inbox is always open.
GitHub
github.com/viet-in-tech