There is a newer version of this site at ayokumo.github.io.
Data Science & Analytics in Sports

Ayokunmi Lawal

Data science student in Toronto, mostly working on football.

Say hello lawalayokunmi25@gmail.com
Ayokunmi Lawal

AL

About

I'm Ayokunmi, a data science student at York University. I like pointing data at football (recruitment models, players the market might be undervaluing, why some players work in some systems), and I also use it on other problems, like housing and public health.

I work in Python, SQL, R and dbt. I like turning analysis into something people can open and use, like a dashboard, a map or a report. I publish the work in public with dates on it, so you can check it yourself.

A few things I've built:

I also reach out to clubs with free analysis built for that club.


Projects

In progress

Football Player Undervaluation Model with RAG Scouting Assistant

I'm testing whether players who moved from nine leagues outside the top five into top five clubs were undervalued at the time, using only what was known before each transfer. The data scope and the assistant scaffold are in the repo. There are no model results or benchmark scores yet, and I won't post any until they're real.

Python, SQL. Building with scikit-learn, PyTorch, LangChain, LangGraph, PostgreSQL, Docker, AWS.
Running

Premier League Match Prediction Pipeline

A pipeline on GitHub Actions that pulls fixtures, form and standings and scores upcoming matches with a simple weighted model I can explain. A trained model checked across past seasons is next. There are no accuracy numbers yet.

Python, GitHub Actions, REST APIs. Building with PySpark and scikit-learn.
In progress

Toronto Housing and Shelter Forecasting Pipeline

I wanted to know if Bank of Canada rate changes show up in Toronto shelter demand. I built a tested dbt and DuckDB warehouse from five public sources (433K row fact table, 26 automated tests). Occupancy rose about 0.64pp for every 1% rate increase, roughly six months later. The univariate SARIMA and Prophet forecasts did not beat a naive baseline, so I'm adding one that uses the rate.

dbt Core, DuckDB, Python, statsmodels.
In progress

Toronto Shelter Overdose Spatial Analysis

A companion to the housing project. I looked at what predicts where overdoses happen across Toronto's 158 neighbourhoods. Low income prevalence explained most of the apparent clustering, and a machine learning layer cut cross validated error from about 52 to about 33 incidents. A hotspot analysis is next.

Python, geopandas, PySAL, scikit-learn.

Experience

Aug 2026 to Present

Sports Analyst Intern

York University

Queried varsity football databases with SQL and Python and wrote R code for 2 weekly coaching reports used in game preparation.

May 2024 to Aug 2024

Financial Data Analyst Intern

Lekki Cooperative Multipurpose Society

Automated recurring Power BI reports on cryptocurrency rate movement tracked with SQL, saving 40 hours per month and keeping investors and senior leadership current.

2026 to Present

Finance Executive

York University Data Science Club

Reconciled purchases and reimbursements and prepared event budgets for a club of 120+ members, reporting discrepancies to the Head of Finance.


Education & Objective

Expected 2028

York University

Bachelor's Degree, Data Science

Available for a 4 month work term from January 2027. Coursework in statistics, machine learning and data engineering, which I apply to football through independent scouting and analytics projects outside the classroom.

Longer term

Sporting Director

Objective

I'm working toward a recruitment or analytics role inside a club. I'm starting with free outreach to local semi professional clubs and academies, using analysis built for each club as the introduction.


Certificates

Completed

Machine Learning Methods and Tools

IBM SkillsBuild

View badge on Credly

In progress

Foundations in Building Machine Learning Models

IBM SkillsBuild
In progress

Machine Learning for Data Science Projects

IBM SkillsBuild