I am a data scientist and engineer with a Master of Data Science from the University of Virginia (UVA MSDS ’25).

I build practical, reliable, and scalable data systems—combining machine learning, cloud tools, automation patterns, and modern AI techniques.

My recent work includes a year-long collaboration with the U.S. Census Bureau, developing a large-scale metadata extraction and document-classification pipeline to improve public data usability.

I bring a unique combination of:

• Strong technical execution
• Real project delivery experience
• Professional discipline from 20+ years in financial services
• High energy, curiosity, and continuous learning

I am open to roles in Data Science, Machine Learning, AI Engineering, Data Engineering, Cloud/DataOps, and GovTech.

Tech: Python · Web Crawling · Metadata Extraction · Regex · NLP · LLMs · Automation

• Designed a recursive web crawler (10-level depth) to map and analyze 3,532 Census Bureau pages.
• Extracted metadata from PDFs, CSVs, XLS/XLSX files using parsing and classification pipelines.
• Built a hybrid rule-based + LLM classification workflow.
• Delivered structured datasets and documentation to Census research leads.

Repository: (will be linked to your capstone repo)

Tech: Python · TensorFlow · Keras · Time-Series Modeling · Explainability

• Modeled patient sequences using LSTM for mortality-risk prediction after cardiac surgery (MIMIC-IV).
• Achieved high accuracy (~98.6%) on LSTM, outperforming baseline RNN/GRU.
• Improving the project with SHAP explainability.
• Rebuilding the pipeline for cleaner engineering and GitHub readiness.

Repository: (link when ready)

Tech: Stan · CmdStanPy · MCMC · Variational Inference

• Implemented stochastic volatility and flight price Bayesian models.
• Conducted posterior diagnostics, trace plots, and correlation analysis.
• Compared MCMC vs. Variational Inference behaviors.

Repository: (add link when ready)

Tech: CNN · Text Classification · Generative Models (DCGAN)

• Built image classification models (CNN).
• Implemented sentiment analysis on IMDB reviews (Bidirectional LSTM, ~86% accuracy).
• Developed generative models using DCGAN.
• Practiced modular deep learning across data, modeling, and evaluation.

Repository: (add links to codeathon repos)

Python · R · SQL · Bash · Git/GitHub · Linux · Jupyter · VS Code · Makefile · GitHub Actions

Neural Networks · CNN · LSTM · GRU · GAN · scikit-learn · TensorFlow · Keras
NLP pipelines · spaCy · Regex · Prompt Engineering · LLM-assisted classification
Generative AI · Explainability (SHAP)

Bayesian inference (Stan, CmdStanPy)
MCMC · Variational Inference
Regression · Probability · Hypothesis Testing · Time Series

Web scraping · BeautifulSoup · Requests
Large-scale crawling (DFS/BFS hybrid)
Metadata extraction (PDF/CSV/Excel)
Cron jobs · Automation patterns · Prototype AWS S3 workflows

• Strengthening MLOps and cloud deployment fundamentals.
• Enhancing the ICU LSTM model with SHAP explainability.
• Expanding the Census metadata system into a more automated pipeline.
• Building clean, recruiter-ready repositories for my portfolio.

• Email: 📧 chiqvdo@gmail.com
• GitHub: 🔗 https://github.com/ChiQuynhDo