What I Enjoy Building
Systems that connect data to real decisions
I'm drawn to projects that have both technical depth and visible usefulness: retrieval systems, pipelines, analytics tools, and data products with a clear downstream user.
Seattle-based | CS + Data Science at UW
Kevin Bui | Data Science & Machine Learning Intern
I like working at the point where engineering, analysis, and product usefulness meet: MCP services, geospatial pipelines, LLM systems, and dashboards that help people make clearer decisions.
I'm most energized by projects that start with raw, messy, real-world data and end with something useful, whether that's a cleaner pipeline, a smarter model workflow, or a visualization that makes the answer obvious.
At NextGen Federal Systems, I build Model Context Protocol services that ground LLMs with structured satellite, weather, and geospatial data, and I've been exploring how to make retrieval systems more secure and more dependable for sensitive document analysis.
At UW, I also support research operations by cleaning and coordinating data for a 500+ participant subject pool. I like the mix of engineering discipline and human usefulness that comes with that work: the data has to be right, but it also has to help someone else do their job better.
I tend to enjoy the in-between work: cleaning the weird dataset, connecting the API no one wants to touch, or translating technical output into something clear enough that another person can act on it.
What I Enjoy Building
I'm drawn to projects that have both technical depth and visible usefulness: retrieval systems, pipelines, analytics tools, and data products with a clear downstream user.
How I Work
I like understanding the full path from source data to final output. Usually that means asking a lot of questions, cleaning edge cases early, and trying to make the system understandable for the next person too.
Outside the Portfolio
The bowling dashboard is probably the most personal project here. It's part sports tracker, part data toy, and part excuse to keep building visualizations around something I care about.
NextGen Federal Systems | Morgantown, WV
University of Washington | Seattle, WA
Designed and deployed 3+ Model Context Protocol services to augment LLM workflows with structured satellite, weather, and geospatial data. Deterministic tool-based retrieval improves model grounding and reduces hallucinations on specialized queries.
Developed and evaluated a secure Retrieval-Augmented Generation system for 50MB+ confidential documents. Implemented TOC-based semantic chunking and embedding-driven retrieval to increase contextual precision in LLM-based analysis tasks.
Built a production-ready pipeline integrating 10+ configurable API parameters for the NOAA GOES satellite feed. Enabled dynamic feature selection and standardized JSON outputs consumed directly by downstream ML analysis.
Engineered an ingestion pipeline processing 1M+ USGS GNIS and NGA GNS records into SQLite, replacing probabilistic API-based lookups with authoritative dataset-driven entity resolution for ambiguous geographic queries.
Analyzed 10K+ transactional records using SQL and Python to identify revenue drivers, customer purchasing patterns, and inventory turnover trends. Designed interactive Tableau dashboards for sales growth, customer retention, and product profitability.
Developed a Python-based data pipeline to scrape, clean, and structure 1,000+ Pokemon records, transforming unstructured HTML into analysis-ready datasets. Performed data cleaning, normalization, and feature extraction with Pandas; results stored in MongoDB.
Personal performance tracker built with Chart.js and Firebase. Tracks scratch scores, strikes, spares, and ball speed across every game, with rolling averages, personal records, and an interactive live scorecard with real bowling scoring logic.
I'm looking for Summer 2026 internships and new-grad roles in data science and ML. If you're building something thoughtful around data pipelines, LLM systems, analytics, or geospatial tools, I'd love to hear about it.
Send a Message