Data preparation
Clean, structure, and validate data so the analysis starts with a dependable foundation.
DATA ANALYST · WASHINGTON, D.C.
I’m Joseph Nwankwo. I bring an analytical mind and an operations perspective to the questions that matter to your business.
01 / ABOUT ME
My background in business management shapes how I approach analytics: start with the real problem, then follow the evidence.
I have hands-on experience supporting business operations, improving processes, and working with teams. That experience helps me connect analytical work to the decisions people make every day.
My graduate studies in Data Analytics at The Catholic University of America build on that foundation. I work with Python, SQL, Power BI, and Excel to organize data, explore patterns, and communicate useful insights.
02 / EXPERTISE
I combine technical tools with business context to make analysis clear, useful, and relevant.
Clean, structure, and validate data so the analysis starts with a dependable foundation.
Translate complex information into dashboards and reports that make patterns easier to understand.
Explore operational performance, economic questions, and opportunities for better decisions.
03 / PROJECTS COMPLETED
AI ENGINEERING
The job search as one system: Spark pipeline ingests live postings, pgvector embeddings rank them against my resume, and an AI agent with 8 read/write tools reasons over it all. 171 postings, 862 embedded chunks — capstone graded A (99/100).
MACHINE LEARNING
Where does Zillow’s own valuation model go wrong? 90,275 sales and 58 property features — LightGBM predicted the pricing error at MAE 0.06822, with tax-assessment features carrying the signal.
DATA VISUALIZATION
Where does social traffic actually come from? 10,000 posts analyzed in Python — YouTube takes 60% of views, Instagram leads engagement — explored through an interactive Tableau dashboard.
MACHINE LEARNING
Can shipment data warn of disruptions before they hit? Gradient Boosting reached 73.9% accuracy and 98.5% recall on lifecycle data — weak spots reported honestly.
MACHINE LEARNING
Airlines change prices constantly — can ML predict the fare? XGBoost hit R² 0.966 on 245,955 records from 1993–2024, beating three other models.
MACHINE LEARNING
Banks waste marketing on customers who will never say yes. On 5,000 records (9.6% acceptance), a bagging ensemble hit 99% accuracy — so campaigns target the right 10%.
DATABASE SYSTEMS
Disconnected records slow hospitals down. Full operations system on Oracle APEX — patients, doctors, appointments, billing — with role-based access.
WEB DEVELOPMENT
A used-car seller needed inquiries, not just listings. Live site with filterable inventory and one-tap WhatsApp contact — stock updates are one edit, auto-deployed.
WEB DEVELOPMENT
Version one of this portfolio: hand-built HTML/CSS/JS. It taught me shipping beats polishing — its DNA is in the site you are reading now.
PYTHON BASICS
Three lines of Python — a=7, m=9, print(a+m) — that grew into pandas pipelines and 10,000-row analyses. Everyone starts somewhere.
MASTER’S DEGREE
The Catholic University of America
Washington, D.C., United States · Sept 2024 – May 2026MASTER’S DEGREE
Mykolas Romeris University
Vilnius, Lithuania · Feb 2018 – July 2020BACHELOR’S DEGREE
University of Maiduguri
Maiduguri, Nigeria · Apr 2010 – Jun 201505 / PERSPECTIVE
171 live postings, 862 embedded chunks, an AI agent with 8 read/write tools — the whole job search as one system. Graded A (99/100).
06 / LET’S CONNECT
For analytics opportunities, collaborations, or a conversation about data, get in touch.
Email JosephBased in Washington, D.C.