Olamide Olayinka
Data Science | Machine Learning | SQL & Python | Risk Analytics | Data Visualization
Data Analytics Portfolio
📍 Ottawa, ON
💼 LinkedIn
📊 GitHub
📧 Email
Education
- IBM Data Science Professional Certificate, Coursera (2025)
- IBM Data Analyst Professional Certificate, Coursera (2024)
- BEng, Mechanical Engineering, Carleton University (2017)
About Me
I am a data professional with expertise in data science, fraud detection, AML policy design, and financial crime analytics. I work with databases, Python, and visualization tools to transform raw data into clear, actionable insights. My experience bridges compliance, fraud prevention, and applied data and machine learning projects.
I have developed interactive dashboards, automated transaction monitoring workflows, and applied statistical analysis to real-world financial data. I am driven by solving problems that safeguard people and organizations, using data as the link between risk and strategy.
Skills
- Programming & Analysis: Python (Pandas, Scikit-Learn, Plotly/Dash, NumPy, Jupyter Notbooks), SQL (MySQL, IBM Db2)
- Visualization & Reporting: Excel, Dash, Plotly Express, Matplotlib, IBM Cognos Analytics
- Data Tools: VS Code, MySQL, Excel.
- Fraud & Compliance: Verafin, LexisNexis ThreatMetrix, Falcon, VCas, TSYS
- Other: HTML/CSS, AML/CTF policy design, regulatory compliance
Projects
💳 Credit Card Fraud Detection (Machine Learning)
Supervised machine learning project identifying fraudulent transactions within a highly imbalanced dataset.
- Processed 284K+ anonymized transactions with PCA-transformed features
- Scaled only Time (RobustScaler) and Amount (StandardScaler) to preserve PCA integrity
- Balanced data using SMOTE oversampling and Random Undersampling to improve minority-class representation
- Trained and evaluated models (Logistic Regression, Random Forest, SVM, Decision Tree) across imbalanced and balanced datasets
- Observed that accuracy decreased on both balanced experiments compared to the imbalanced dataset — a natural result of reducing the dominance of the majority (non-fraud) class; however, recall and ROC AUC significantly improved, indicating a stronger ability to detect actual fraud
- The Undersampled balanced Logistic Regression and Linear SVC delivered consistently high recall and ROC-AUC without overfitting, making them more reliable for real-world use.
- Visualized and interpreted model performance using ROC and confusion matrices
📊 Ottawa Shooting Incidents (2018–2024)
Interactive Dash app with MySQL backend analyzing firearm-related incidents by time, injury severity, and neighborhood.
- Built ETL pipeline to clean and load city data
- Developed a 2x2 grid dashboard with stacked bar and line charts
- Added dynamic filtering for injury severity and year of incident
🚀 SpaceX Falcon 9 First-Stage Landing Prediction
Machine learning project predicting the success of Falcon 9 booster landings using launch and payload data.
- Collected and cleaned launch data using Python and SQL
- Performed exploratory data analysis to identify key drivers of landing success
- Built and compared models (Logistic Regression, Decision Tree, SVM) with ROC-AUC evaluation
- Developed interactive Plotly Dash dashboard to visualize launch outcomes by payload, orbit, and site
Work Experience
Compliance Associate — Renno & Co. (2025)
Fraud Data Specialist — Alterna (2021–2025)
Fraud Detection Analyst — TD (2019–2020)