About Me
Analytical by nature,
data-driven by practice.
I'm a Data Scientist and Business Analyst based in Nairobi, Kenya, with over three years of progressive experience at Kenya Commercial Bank (KCB). My work spans credit risk analysis, financial forecasting, KPI reporting, and machine learning model development. I hold a Bachelor of Commerce in Accounting & Finance from Dedan Kimathi University of Technology and am currently pursuing an MSc in Business Analytics. I'm proficient in Python, SQL, Power BI, and the Scikit-learn ML stack — and I'm passionate about using data to drive decisions that are both commercially sound and socially meaningful.
Background
Experience & Education
Work Experience
Bank Officer
Oct 2024 – Present
Kenya Commercial Bank (KCB) · Retail Banking Division
- Monitored and managed credit portfolio health — tracking NPLs, overdrawn accounts, and liquidation accounts, and generating reconciliation reports and credit returns within defined turnaround times
- Analysed force pay reports and loan callback data to identify credit risks, recommending actions and escalating findings to line management
- Processed and reconciled branch transactions and account operative tools, ensuring data accuracy and integrity across teller and customer service operations
- Reviewed agent and merchant compliance against bank and regulatory requirements, maintaining accurate channel performance records
- Applied AML and KYC procedures during daily operations — screening transactions, resolving flagged customer instructions, and escalating suspicious activity as required
- Implemented service improvement initiatives at branch level and provided structured feedback to management based on operational data
Bank Officer Trainee
Oct 2023 – Oct 2024
Kenya Commercial Bank (KCB) · Retail Banking Division
- Received, processed, and reconciled cash instruments and customer transactions within required TATs, maintaining accurate teller and account records
- Processed customer requests including account openings, balance certificates, and correspondence — capturing and resolving feedback via eCRM within defined turnaround times
- Reviewed wallet share data to identify cross-sell opportunities, engaging customers and contributing to branch revenue collection targets
- Coordinated banking hall activities and supported implementation of service improvement initiatives under the guidance of the Branch Manager
- Adhered to AML, KYC, and Business Continuity procedures in all daily operations, contributing to risk mitigation and regulatory compliance
Retail Direct Sales Representative
Mar 2022 – Oct 2023
Kenya Commercial Bank (KCB)
- Delivered sales targets across Retail, Agri, SAHL, Agent Banking, and Bancassurance product lines, preparing regular performance reports and trend analyses for business planning
- Collected and synthesised customer feedback data to evaluate product performance and support campaign targeting decisions
Graduate Assistant – Research & Data Analysis
Jun 2019 – Dec 2021
Dedan Kimathi University of Technology
- Assisted lecturers in preparing teaching materials and facilitated tutorials and discussion groups for ~500 students across business and finance programmes
- Supported student research projects — data collection, quantitative analysis, and academic report preparation
- Tutored Financial Management, Quantitative Methods, Financial Statement Analysis, Taxation, and Company Law
Education
Data Science, Analytics & AI Bootcamp
Graduating June 2026
January 2026 – June 2026
LuxDev HQ
Intensive 6-month programme covering machine learning, AI applications, advanced analytics, and enterprise data platforms. Business intelligence using Excel, Power BI, SQL, and Python.
MSc Business Analytics
In Progress
2019 – Present
Dedan Kimathi University of Technology
Bachelor of Commerce – Accounting & Finance
2015 – 2018
Dedan Kimathi University of Technology
CPA Part I
2016 – 2017
KASNEB
Learning & Development
Python ML Projects
1 · Regression · Supervised Learning
Linear Regression Basics
Implements Simple Linear Regression using Python and Scikit-learn. Covers data exploration, EDA, train-test split, model training, and evaluation using MSE and R² Score. Predicts house prices from a continuous variable.
View on GitHub →
2 · Regression · Multiple Features
Multivariate Linear Regression
Extends simple regression by incorporating multiple predictors. Demonstrates how different features collectively influence a target variable, with applications in financial forecasting and customer behaviour analysis.
View on GitHub →
3 · Feature Engineering · Encoding
Dummy Variables & One-Hot Encoding
Demonstrates how to handle categorical variables using dummy variables and one-hot encoding — essential preprocessing techniques for ML models. Covers the distinction between encoding strategies and the dummy variable trap.
View on GitHub →
4 · ML Pipeline · Data Splitting
Train-Test Split for Machine Learning
A practical guide to correctly dividing datasets into training and testing sets. Covers data leakage prevention, reproducibility via random state, and split ratios — using car price data as a real-world example.
View on GitHub →
5 · Classification · Binary
Logistic Regression – Binary Classification
Implements Logistic Regression for binary classification using insurance data. Covers the sigmoid function, decision boundary, probability-based prediction, and model evaluation with accuracy and confusion matrix.
View on GitHub →
6 · Classification · Multiclass
Logistic Regression – Multiclass (Digits)
Extends Logistic Regression to classify handwritten digits (0–9) using Scikit-learn's digits dataset (1,797 samples, 64 pixel features). Applies One-vs-Rest strategy and evaluates accuracy and confusion matrix.
View on GitHub →
7 · Classification · Tree-Based
Decision Tree – Salary Prediction
Applies a Decision Tree Classifier to predict whether an employee earns above $100K based on company, job role, and degree. Demonstrates categorical encoding, interpretable rule-based logic, and HR data analysis.
View on GitHub →
8 · Classification · Margin-Based
Support Vector Machine (SVM) – Iris Classification
Implements SVM to classify Iris flowers into 3 species. Covers hyperplane optimisation, One-vs-Rest multiclass strategy, kernel concepts, and model evaluation on the Iris dataset.
View on GitHub →
9 · Ensemble Learning · Classification
Random Forest – Handwritten Digit Recognition
Uses a Random Forest Classifier on the Digits dataset. Demonstrates bootstrap sampling, majority voting, variance reduction over single trees, and feature importance extraction.
View on GitHub →
10 · Model Validation · Cross-Validation
K-Fold Cross Validation – Digits Classification
Demonstrates K-Fold Cross Validation to provide a robust model performance estimate beyond a single train-test split. Shows how average accuracy across folds reduces variance and detects overfitting.
View on GitHub →
11 · Unsupervised Learning · Clustering
K-Means Clustering – Customer Segmentation
Applies K-Means Clustering to group individuals by age and income. Uses the Elbow Method to determine optimal K and visualises segments — demonstrating real-world customer profiling and market segmentation.
View on GitHub →
12 · Probabilistic · NLP
Naïve Bayes – Titanic Survival & Spam Detection
Implements Naïve Bayes on two problems: predicting Titanic survival from structured passenger data, and classifying email spam vs. ham using text vectorisation. Demonstrates probabilistic ML on both structured and unstructured data.
View on GitHub →
13 · Model Optimisation · Hyperparameter Tuning
Hyperparameter Tuning & Model Selection – Iris Classification
Compares Train-Test Split, K-Fold Cross Validation, and GridSearchCV on the Iris dataset. SVM with RBF kernel achieved ~98% CV accuracy and was selected as the optimal model through exhaustive automated search, outperforming Logistic Regression and Decision Tree baselines.
View on GitHub →
14 · Regression · Regularisation
L1 & L2 Regularisation – Lasso & Ridge Regression
Applies Lasso (L1) and Ridge (L2) regularisation to the Melbourne Housing Dataset to address overfitting in high-dimensional data. Covers data cleaning, one-hot encoding, correlation analysis, and model comparison.
View on GitHub →
15 · Classification · Instance-Based Learning
K-Nearest Neighbors (KNN) – Iris Classification
Implements KNN classification on the Iris dataset using distance-based majority voting. Covers feature scaling, optimal K selection, and model evaluation — highlighting the trade-off between small K (noise-sensitive) and large K (underfitting).
View on GitHub →
16 · Dimensionality Reduction · Unsupervised
Principal Component Analysis (PCA) – Digits Dataset
Applies PCA to the Digits dataset (64 features, 10 classes) across three scenarios: full 64 components (97.22% accuracy), 95% variance retained (96.95%), and 2 components (60.80%). Demonstrates that retaining 95% variance preserves nearly all predictive power while significantly reducing dimensionality.
View on GitHub →
Portfolio
SQL Projects
Advanced SQL · Banking
Credit Portfolio Performance Dashboard (SQL Layer)
Designed complex multi-table SQL queries using CTEs and window functions to extract NPL trends, overdrawn account patterns, and loan liquidation metrics from a relational banking database.
View on GitHub →
Data Analysis · Reporting
Agent Banking Channel Analytics
Wrote parameterised SQL queries to analyse transaction volumes, revenue collection, and agent performance across multiple banking channels, feeding downstream Power BI dashboards.
View on GitHub →
Data Quality · ETL
Data Validation & Reconciliation Scripts
Built a suite of SQL scripts to automate data quality checks — detecting duplicates, mismatched records, and null values — used for pre-submission regulatory report validation.
View on GitHub →
Add SQL Project
Portfolio
Power BI & Excel Projects
Power BI · HR Analytics · People Management
HR Analytics Dashboard – Full Workforce Insights
An interactive Power BI dashboard delivering end-to-end HR analytics across 1,470 employees. Tracks gender distribution (60% male / 40% female), promotion eligibility (5% due), retrenchment status (117 flagged), job level headcount, years of service distribution, and employee distance from work — all filterable by Department and Education Field.
View on GitHub →
Excel · Data Cleaning · Real Estate
Real Estate Price Prediction – Data Cleaning (Excel)
A comprehensive data cleaning project using Excel on a real estate dataset. Covers handling missing values, removing duplicates, standardising formats, and preparing structured data for downstream predictive modelling and visualisation.
View on GitHub →
Power BI · Predictive Analytics · Real Estate
Real Estate Price Prediction – Power BI Dashboard
An interactive Power BI dashboard visualising real estate price prediction insights. Displays key pricing drivers, property distribution, and predictive trends — translating cleaned data into actionable business intelligence for property market analysis.
View on GitHub →
Add Power BI / Excel Project