Prince Okunade.
PRINCE OKUNADE / DATA, AI & MACHINE LEARNINGBased in London

Hello, I'm Prince

Understand your data.
Predict what matters.
Build useful AI.

I’m Prince, a London-based data scientist specialising in machine learning and AI engineering. I help businesses turn complex data into clear decisions, anticipate outcomes and build AI tools that make everyday work easier. With a foundation in Statistics, I take work from defining the problem and preparing the data through to testing and deploying a usable solution.

Evaluated RAG · Cloud-deployed ML · Published research
BSc Statistics · MSc Data Science, York St John University

Open to work · Data Science · Machine Learning · AI Engineering

Explore my work Let's talk
PRINCE OKUNADE
01 / THE PERSON

A life connected
by curiosity.

Numbers are part of my story.
People give them a purpose.

My journey has taken me from studying Statistics at the University of Ibadan to studying Data Science at York St John University in London. Along the way, I've worked with agricultural data, taught mathematics and helped take financial literacy into communities.

FOUNDATIONBSc Statistics

University of Ibadan · 2019–2024

CURRENT CHAPTERMSc Data Science

York St John University · London

02 / SELECTED WORK

Questions worth
working on.

Browse code on GitHub

Each project starts from a decision someone needs to make. Here is why I built it, how I built it and what it changes. The case studies hold the full evidence, and the demos and code are open to inspect.

01 / FEATURED PROJECT · APPLIED AI

Grounded Drug-Safety Assistant

A drug-safety assistant is only useful if people can trust it, and a general chatbot will answer confidently even when it has no source. I built one that answers only from UK MHRA safety documents, cites the passage behind every answer and refuses when the documents do not cover the question. I tested every upgrade before keeping it: combining keyword and meaning-based search made answers more reliable, and removing a component that slowed responses without changing any answer made it fast enough to use.

The effectAn AI assistant that says “I don’t know” instead of inventing an answer, which is what makes it safe to rely on in a regulated setting.
Read the full case study
Python · FAISS · BM25 · Claude · FastAPI · Docker
Read the articleView code on GitHubTry the demo Cached examples
02 / ML ENGINEERING

Churn Prediction as a Service

A retention team cannot call every customer, so the real question is who to call first. I ranked customers by their risk of leaving and used SHAP to show why each one was flagged, so the team can act on the reason, not just the score. I deployed it on Google Cloud as a live service that redeploys automatically and warns when predictions drift away from the pattern it was trained on.

The effectCalls go to the customers most likely to leave, and the team can see when the model needs retraining instead of trusting it blindly.
Read the full case study
XGBoost · SHAP · FastAPI · Cloud Run · BigQuery · CI/CD
View code on GitHubTry the demo Instant results preview
03 / DATA ENGINEERING & ANALYTICS

London Smart-Meter Lakehouse

Raw smart-meter readings are too messy to compare households, weather and tariffs consistently. I built a governed pipeline on Azure Databricks that turns half-hourly readings into clean, analysis-ready tables, then grouped households by the shape of their daily demand and forecast consumption from weather and calendar data. When the cloud setup blocked Spark’s machine-learning library, I kept the large-scale preparation in Spark and moved the modelling to scikit-learn rather than dropping the pipeline.

The effectA clear view of when and how London households use electricity, the starting point for demand planning and tariff analysis.
Read the full case study
PySpark · Databricks · ADLS Gen2 · Delta Lake · SQL · Power BI
View code on GitHub
FROM THE RAG PROJECTI measured every obvious upgrade in my RAG. Most of them lost.Read the accompanying article on Medium

More explorations

Stock direction modellingTested whether popular trading indicators really predict next-day moves by checking three models only on days they had never seen. None beat simply guessing “up” every day, and the most-tuned model looked best in training but failed on new data: the kind of result that stops money being bet on a false signal.World Cup forecasting A single match prediction does not tell you a team’s chance of winning a tournament. I turned match probabilities into 10,000 simulated tournaments, so the forecast shows each team’s chance of reaching every stage instead of naming one confident winner.Life expectancy modelling Group coursework comparing regression models on country health and socioeconomic data. The near-perfect scores were the warning sign: I traced them to the same countries appearing in both training and testing, and set out how to evaluate it properly.

Tools I work with

From analysis and model evaluation to APIs, cloud deployment and monitoring.

Data science & analytics

PythonRSQLPandasNumPyPower BIMatplotlib

Machine learning & AI

scikit-learnTensorFlowXGBoostRAGLLM evaluation

Engineering & deployment

DockerFastAPIGitGitHub ActionsGoogle CloudAzure MLBigQuery
03 / EXPERIENCE

Working with data.
Working with people.

JAN–SEP 2026LONDON

Mathematics Tutor (GCSE)

Get Further · Waltham Forest College

I taught 20+ students resitting GCSE mathematics, identifying where their reasoning broke down and adapting explanations to help them work through difficult concepts.

The role sharpened my communication: understanding the question someone is really asking, checking their understanding and finding a useful way to explain the answer.

GCSE mathematics · Problem-solving · Communication
SEP 2024–AUG 2025ABEOKUTA

Mathematics Teacher (NYSC)

The Cross College

I taught Mathematics and Further Mathematics to 100+ students across JSS1 to SS3, covering topics including probability, calculus and statistics.

I reviewed the curriculum with teachers and school leadership, presented recommendations for different year groups and coordinated fellow NYSC teachers.

Mathematics · Curriculum review · Team coordination
JUN–SEP 2023IBADAN

Data Science Intern, Biometrics Unit

International Institute of Tropical Agriculture (IITA)

I cleaned and validated 15,000+ observations from 8+ agricultural trial sites in R, resolving recording inconsistencies between sites to create an analysis-ready dataset.

I fitted regression and predictive models to support agricultural research, then presented the findings and their limitations to senior stakeholders, including heads of unit and the head of data.

Data quality · R · Statistical modelling · Stakeholder communication
NYSC COMMUNITY LEADERSHIPABEOKUTA NORTH

President, Financial Inclusion CDS

National Youth Service Corps

Led financial literacy outreach across more than 10 schools, markets and communities, helping bring practical financial education to wider audiences.

Coordinated people and outreach activities around a shared public-service goal.

Leadership · Financial literacy · Community outreach
04 / PUBLISHED RESEARCH
PEER-REVIEWED · CO-AUTHOR

Understanding outcomes.
Accounting for uncertainty.

Bayesian Analysis of COVID-19 Recovery and Death Cases in Nigeria

Research rooted in my undergraduate work, using Bayesian survival analysis to study recovery and death outcomes in Nigerian COVID-19 data.

Journal of Scientific Research and ReportsRead the paper
05 / BEYOND THE WORK

Outside the projects.

I love sport, from following the competition to exploring its uncertainty through my World Cup forecasting project. Hospitality work has also taken me into London's stadiums and event venues.

I contribute behind the scenes in church media and have volunteered as a mental health ambassador with the Asido Campus Network. Teaching, community work and team settings have all shaped how I listen and explain.

London cityscape
06 / SAY HELLO

Good things start
with a conversation.

I’m open to opportunities in data science, machine learning and AI engineering, and to collaborations where I can put data and AI to practical use.

Discuss a roleExplore a collaboration
LinkedIn GitHub