Data scientist & researcher
I'm Cynthia, a mathematician turned data scientist finishing my MS at GVSU. I build the pipelines, models, and dashboards behind decisions that actually reach people: girls in classrooms, international students chasing visas, communities that rarely get counted. Rigorous about the work, genuinely delighted by the hard parts.
(and yes: DAX measures appear in my dreams. I've made peace with it.)
The why
Who is being missed, and what would it take to reach them?
Where it started
I grew up watching bright kids quietly disappear from school for reasons no one was counting. Sometimes the cause was as ordinary as a missing pack of sanitary towels: without them, girls missed days, then weeks, then stopped coming back at all. I wanted to fix that long before I knew what a regression was.
Years later, the data gave me a way. In a community enrollment campaign, the numbers made the invisible visible, and turned quiet concern into action. That was the moment I understood what analysis was actually for.
When it became the work
At a national curriculum institute, that question became my full-time job: student data at scale, special-needs fieldwork, analysis that shaped real curriculum and inclusion decisions.
For the first time, my work was changing outcomes for thousands of students I would never meet. That is still the whole point of all of it, and every project since has been a version of the same question.
Selected work
Projects I built myself, where I can open up the data, the methods, and the messy parts.
Machine learning / Clinical
A full Knowledge Discovery pipeline on 2,149 patients reaching 93.8% accuracy with a decision tree you can actually read. When clustering failed, I reframed it as a finding about disease continuity, then proved it.
Pythonscikit-learnmlxtend
Read the case →Statistical modeling / Policy
25,391 students across a family of Generalized Linear Models: family engagement gives an 18% stronger protective effect for low-income students (p = 0.009), replicated across four models, with direct implications for targeted interventions.
RtidymodelsGLMs
Read the case →Statistical inference
Merged Census ACS data with 100,000+ gun-violence incidents across 50 states. Permutation tests, a 10,000-sample bootstrap, and Welch's t-test showed poverty predicts gun-violence rates; population does not.
RQuartoACS
Read the case →Data engineering
A production pipeline moving 96,000+ records through a five-phase workflow: extraction, transformation, dual-target loading to SQLite and BigQuery, validation, and a self-service Looker dashboard.
PythonSQLBigQuery
Read the case →Agentic AI / Immigration
An agentic system grounded in real USCIS H-1B data that helps international students find employers who actually sponsor visas. I'm building it because I needed it; when an early version gave insensitive advice, I caught it and fixed that too.
PythonLLM agentsUSCIS data
Try the live app →Research & professional work
Studies and roles with research labs and employers. I can share my role and the kind of work, but the data and methods belong to the project, so these stay deliberately high-level. Happy to talk through any of them directly.
GVSU / Blueprint Labs (MIT)
My role: data scientist on the research team
Contributed to a validation study modeling educator burnout and its drivers on real charter-school data, using latent-profile and regression methods. The dataset and approach belong to the project, so I keep the specifics offline; I'm glad to walk through my contribution in conversation.
National curriculum institute
My role: data analyst & researcher
Situational analysis of student-to-textbook ratios to optimize allocation across schools, plus data processing for ICT and Special Needs Education integration and national curriculum-transition work. Government data, so this stays high-level.
How I work
I engineer
Production ETL across 34K+ students and 96K+ records: validated, logged, and version-controlled, not held together with hope.
I model
Interpretable models replicated across independent methods. If I can't explain why it's true, I don't ship it.
I translate
Built for parents, teachers, and policymakers, not just other data people. The insight only counts if someone can act on it.
I'm a researcher, not just an analyst: I design and validate systems, I don't just report what happened. I build with modern tools, but I own every line; I ship what I understand.
The path
A community enrollment drive taught me that the most important variable in any dataset is usually a person. I've never unlearned it.
I came for the rigor of mathematics and stayed for what it could do in the world. Volunteer tutor on the side.
National student data and special-needs fieldwork feeding real curriculum policy; 15+ analytical reports informing a national curriculum transition.
Funded through a Blueprint Labs grant. Deepening the statistics, machine learning, and data engineering I now put straight into production.
One assistantship, two chapters across two departments.
Beyond the data
Leadership isn't a line on a résumé for me, it's standing rooms, real budgets, and students who need someone in their corner.
Off the clock
When I'm not in the data: yoga to think straight, anime for the stories, and adrenaline for the rest. I'm scared of heights and have done it anyway (early results suggest this pattern will continue). My bucket list has The Amazing Race on it and all seven continents; I'm three down so far. Faith keeps me grounded through all of it.
Let's work together
Data Analyst, Data Scientist, Analytics Engineer, and Data Engineer roles. Open to industries, open to relocation, and ready when a serious company is.
Yes, I'll need visa sponsorship; I've literally built tools to navigate it, so I know the process cold.
Helping people is the point. The rest is byproduct.