WorkUser Activity Pipeline
User Activity Pipeline
Personal project - Solo: concept, data modeling, notebook implementation, and frontend storytelling.
Overview
A NumPy-first analytics build that starts from raw user events, engineers features, computes engagement scores with matrix math, and presents the output in a product-style dashboard.
I did not want to memorize NumPy operations in isolation and then struggle to apply them later. I wanted one artifact where the data pipeline story stayed intact from source events to scored output.
The notebook plays the role of a junior data engineer receiving a raw export, transforming it into model-ready features, and producing engagement signals that could feed a product surface. The dashboard then takes that same output and presents it as if it were part of a live analytics workspace.
What it covers
- Synthetic operational-style dataset with explicit transformation stages.
- Feature engineering, normalization, aggregation, and matrix-based scoring in pure NumPy.
- Frontend analytics surface that visualizes patterns, score drivers, and row-level user details.
What I owned
- Owned the full narrative from architecture framing to code implementation to visual presentation.
- Prioritized clarity under pressure by keeping each transformation step explainable and testable.
- Used the project as a bridge between learning depth and portfolio communication quality.
Outcomes
- Connects systems thinking (OLTP/OLAP, ETL, derived data) to hands-on implementation instead of treating them as separate study tracks.
- Keeps the scoring model legible with feature-level contribution breakdowns, so numbers can be explained under scrutiny.
- Pairs a technical notebook with an interactive dashboard, making the same pipeline readable to both engineers and non-technical reviewers.
- Documents open technical gaps honestly, which turns the project into a credible learning artifact rather than a polished demo with missing context.
Built with
For reviewers who need the implementation picture: tools and libraries, not the business story above.
- Python
- NumPy
- Jupyter Notebook
- Next.js
- TypeScript
- Tailwind CSS
- Vercel