I spent years building interfaces that displayed data before I started asking harder questions about what the data actually meant. That shift led me from frontend into analytics, data engineering, and now into machine learning — moving up the stack, one question at a time.
I started in frontend — React, TypeScript, building the dashboards and data-heavy UIs that analytics teams relied on daily. It was good work. But somewhere along the way I kept gravitating toward the same question: why do the numbers look the way they do?
That kept happening. I started pulling queries on our analytics stack after hours. Then taught myself Python. Then built a logistic regression model on loan data just to see if I could. And then I couldn't really go back.
The jump to data analytics happened about a year and a half ago. Since then — segmentation models, attrition prediction, CLV forecasting with XGBoost and SHAP. The frontend background helps more than I expected: thinking about how someone reads a finding is the same muscle as thinking about how they use an interface.
Now I'm in the middle of the stack — comfortable writing the pipeline that cleans the data, the model that makes sense of it, and the interface that puts it in front of the person who needs it.
The path wasn't planned. It followed the questions I couldn't stop asking.
Five projects across two disciplines. Each one built to answer a real question, not to tick a skills box.
RFM segmentation with K-Means, XGBoost to predict 90-day customer value, SHAP to explain every prediction individually. Business cost output that answers "what is this worth?" not just "what is its AUC?"
Three models head-to-head. Logistic Regression, Random Forest, Gradient Boosting — threshold optimized for F1, with a dollar-value cost analysis that frames the model's value in business terms, not just metrics.
Full EDA on 255K loan records. Logistic regression baseline with a proper scikit-learn pipeline, stratified CV, and modular src/ layout. The project where it all started — structured from day one like something meant to be maintained.
Production-grade analytics dashboard in React 18 + TypeScript. Zustand for state, Recharts for charts, custom hooks, skeleton loading, compare periods, sortable tables. Written for real users, not a tutorial.
dnd-kit drag and drop (pointer + keyboard), Zustand with localStorage persistence, React Hook Form + Zod validation, full ARIA keyboard navigation. Discriminated union types for cards and modal state throughout.
More on GitHub
All repos are public with full commit histories, READMEs with architecture notes, and clean modular code.
github.com/GOLDENIG37Honest self-assessment. Green is where I'm strongest, teal is data engineering foundations, coral is where the journey started.
Three different titles, one common thread: using data to make decisions that actually get made.
EDA, dashboards, cohort analysis, RFM segmentation, business cost framing. I can pull the data, find the signal, and present findings in a way that makes the decision obvious — without losing non-technical stakeholders in the methodology.
Building pipelines that make analysis possible. ETL design, data modelling, SQL, dbt. The frontend background means I actually care about what downstream users need from the data — not just whether the job ran green.
Actively building toward ML engineering. The credit risk, attrition, and CLV projects are the proof of work. Looking for a role where I can grow from strong analyst fundamentals into full model ownership and deployment.
Open to Data Analyst and Data Engineer roles, actively building toward Data Science. If you're working with messy data and need someone who can make sense of it and communicate it clearly — let's talk.