UX Portfolio & Dashboard Suite

Three working data dashboards shipped as products, not screenshots — usable on this page.

Use it — live on this page Source Original demo

Interfaces that outlive frameworks
Notebook analysis to usable product
Work-first information hierarchy

Dashboards: 3 live builds
Open data: 15+ CSVs in-repo
Hosting: GitHub Pages, no backend
Longevity: Framework-free artifacts

Overview

A dashboard suite where the case studies are the artifacts themselves: interactive dashboards for Formula 1 analytics, a Pokedex explorer, and an automobile dataset — each one a self-contained HTML build shipped alongside the raw CSV data that powers it, deployed on GitHub Pages and embedded live on this page.

Problem

Data-analysis portfolios usually reduce weeks of work to a static PNG of a chart. That throws away the thing that makes a dashboard good — the interaction. I wanted visitors to filter, hover, and explore the same views I built during analysis, without installing anything.

Goals

Present analysis work as living dashboards, not screenshots
Keep every dashboard self-contained and shareable as a single URL
Ship the underlying datasets in the open so the work is verifiable

Solution

Each dashboard — Automobile, F1 Analytics, Pokedex — is built as an independent HTML entry point deployed to GitHub Pages, so it loads instantly, cannot break its neighbors, and can be linked or embedded directly (which is exactly how this page serves them).

The repository doubles as the data source: the full Kaggle Formula 1 relational set (circuits, drivers, lap times, pit stops, qualifying, standings — 15+ CSVs) plus the Pokemon and Automobile datasets live in public/, making every visual reproducible.

Process

1Analysis first in Jupyter (the F1, Pokedex, and Automobile notebooks in sibling repos), then translated the strongest views into interactive form
2Each view had to survive the promotion test: only charts that changed a conclusion in the notebook earned a place in the dashboard

Tooling

Tech stack

React
JavaScript
HTML
CSS
GitHub Pages

Friction

Challenges

Dashboards that outlive their framework

Framework-bound dashboards rot when dependencies move. Building each one as a standalone HTML artifact means the F1 dashboard from 2023 still opens in one click today, independent of the React shell around it.

Serving relational data statically

The F1 dataset is a dozen linked tables. With no backend on GitHub Pages, the joins had to happen client-side — which forced deliberate choices about which pre-joined slices each view actually needs.

Takeaways

Lessons learned

Static hosting is a design constraint that improves the design — every feature must justify its data weight
A portfolio that lets the visitor do the analysis is more persuasive than one that describes it
Keeping raw data in the repo turns a portfolio into a reproducible artifact

Roadmap

Future improvements

Responsive redesign of the older dashboard builds — they predate my mobile-first practice
Add the IPL and Game of Thrones work as dashboards four and five
Lighthouse pass on the older dashboard builds

Continue

Analytics

Automobile Data Visualization

Data Science

Formula 1 Data Analysis