[ Iowa City, IA ] — Data Analyst / Applied Data Scientist

John
Kirima

Data Analyst / Applied Data Scientist —
building at the intersection of analytics, machine learning, and decision intelligence.

I build data systems that uncover hidden risk, surface real user behavior, and turn messy information into decisions that matter.

About

I’m John Kirima — a data analyst and applied data scientist focused on turning messy information into clear, useful decisions.

What pulls me to this field is not just modeling or dashboards on their own, but the deeper question underneath them: what is the data actually saying, what is being missed, and what changes once you see it clearly?

My work sits at the intersection of analytics, machine learning, and systems thinking. I’m especially interested in customer behavior, product friction, hidden risk, decision intelligence, and the places where standard tools fail quietly. I like projects that go beyond surface-level reporting and push toward explanation, structure, and business meaning.

That’s the kind of work I’ve been building: systems that audit model blind spots, workflows that automate parts of the data science process, and analyses that turn noisy real-world data into something decision-makers can actually use.

I care about clarity, rigor, and usefulness. Good analysis should not just look technical — it should reveal something important.

Focus: Analytics · ML · Decision Intelligence
Based: Iowa City, IA
Education: Business Analytics & Information Systems,
University of Iowa — Tippie College of Business

What I Work On

I’m most interested in work where data reveals something that standard reporting, baseline models, or surface-level metrics fail to capture.

—customer intelligence
—product analytics
—hidden risk detection
—NLP evaluation
—workflow automation
—decision-focused analytics

Projects

01 — FLAGSHIPCompetitive Sentiment Intelligence

Fintech Sentiment
Intelligence

Why are a quarter of the most critical complaints invisible to the tools companies trust? This case study reverse-engineers sentiment failure modes across the fintech apps people use every day.

A competitive analytics study analyzing 10,386 Google Play reviews across Cash App, Venmo, Chime, and PayPal. I scored sentiment with VADER, validated it against star ratings, and modeled topics with LDA to expose where baseline NLP quietly misreads serious complaints — then translated severity into competitive intelligence and a $2–4M revenue-recovery estimate.

10,386: Reviews analyzed
4: Fintech apps
26%: Critical complaints missed
$2–4M: Revenue recovery est.

hidden complaint detection
sentiment failure modes
severity scoring
competitive intelligence
business impact

Python / VADER / LDA / Pandas / MatplotlibView Case Study↗

DataForge

End-to-end data science is slow, repetitive, and easy to do inconsistently.

A terminal-based tool that automates the full data science workflow through a 9-agent LLM pipeline. Agents coordinate across data ingestion, quality auditing, cleaning, EDA, feature engineering, statistical testing, modeling, SHAP interpretability, and final recommendations — powered by DeepSeek V3.2 and Claude Sonnet.

Python / LLM APIs / DeepSeek / ClaudeView Project↗

Incident Report Analytics Dashboard

Raw incident data sat unused, invisible to the people who needed to act on it.

Applied analytics work building an interactive Power BI reporting system over incident data for a healthcare organization serving individuals with disabilities. Requirements were drawn directly from user interviews, and a SharePoint-to-Power BI pipeline kept the data current without manual handling.

Power BI / SharePoint / SQL

Spotify Track Analysis Application

Track metadata is rich, but flat files make it hard to ask real questions.

A normalized relational schema and web-based analytics application for exploring Spotify track metadata. Advanced SQL surfaces patterns across popularity, energy, duration, and genre over thousands of tracks.

Oracle APEX / SQLView Project↗

Multi-Ownership in European Soccer

Ownership networks shape competition in ways league tables never show.

A study of multi-ownership models across European soccer leagues, joining team performance, financial data, and ownership structures. Tableau dashboards visualize how ownership network effects ripple into competitive outcomes.

Tableau / Kaggle / Data Mining

Personal Landing Page

Most portfolio sites look the same because they start from a template.

A minimal, brutalist site built from scratch in vanilla HTML, CSS, and JavaScript — no templates, no frameworks. Semantic markup, WCAG 2.2 AA compliant.

HTML / CSS / JSView Source↗

Tools / Capabilities

Programming & Databases

Python
SQL
Oracle APEX
HTML / CSS

Data & Visualization

Power BI
Tableau
Microsoft Excel
ETL Processes
Data Cleaning

AI & Machine Learning

Machine Learning
Statistical Analysis
SHAP Interpretability
LLM Integration
Multi-Agent Systems
NLP
BERTopic
VADER
Sentiment Analysis

Applied Work

Role

Data Analyst Intern

Organization

Systems Unlimited, Inc.

Iowa City, IA

During my data analyst internship, I built interactive Power BI reporting systems for incident analysis, automated data flows between SharePoint and Power BI, and worked from stakeholder needs to create tools that improved operational visibility.

Contact

If you’re building interesting things in analytics, product, machine learning, or decision intelligence, I’m always open to connecting.

JohnKirima

Data Analyst / Applied Data Scientist — building at the intersection of analytics, machine learning, and decision intelligence.