About Projects Skills Contact Resume

Hi, my name is

Benedict Bett.

I turn data into decisions.

Data scientist focused on machine learning, predictive modeling, and real-world AI solutions. Currently building intelligent systems that solve meaningful problems across finance, telecoms, and beyond.

84.2% ROC-AUC · Churn Model
307K+ Samples · Credit Model
3+ ML Projects Shipped
View my work Get in touch
scroll
01

About me

I'm an aspiring Data Scientist with a strong passion for extracting meaningful insights from data and building intelligent systems. My interests span across Data Science, Data Analysis, and Machine Learning — where I focus on understanding patterns, solving complex problems, and creating data-driven solutions.

My journey into data began with a fascination for how data shapes decision-making. I'm committed to continuously expanding my skills in statistical analysis, predictive modeling, and emerging technologies in the AI and ML landscape.

I'm actively seeking internship opportunities to apply my knowledge in real-world scenarios and contribute to impactful projects. I believe in the power of data to drive innovation — and I approach every problem with curiosity and dedication.

Technologies I work with:

  • Python
  • scikit-learn
  • TensorFlow
  • Pandas
  • SQL
  • LightGBM
  • SHAP
  • Streamlit
  • Matplotlib / Seaborn
  • Tableau
Benedict Bett
BB
Open to opportunities
02

Projects

Featured Project

M-Pesa Financial Advisor

A RAG-powered personal finance coach for Kenyans. Upload your M-Pesa statement PDF and ask natural language questions about your spending, saving, and sending habits. Answers are cited directly from your real transaction data — no guessing, no generic advice.

RAG Architecture
Live Full-Stack App
2 Knowledge Sources
  • LangChain
  • ChromaDB
  • Groq / Llama
  • FastAPI
  • Next.js
  • HuggingFace
  • Docker
  • Railway
Blog Post Live App Source Code
Spending Breakdown
M-Pesa Statement Food 35% Transport 25% Airtime 20% Other 20%
Featured Project

Safaricom Churn Intelligence

ML-powered customer churn prediction built for Kenya's telco market. Engineered Kenya-specific features including M-Pesa usage patterns, Bonga loyalty points, and county-level demographics to identify at-risk customers before they churn.

84.2% ROC-AUC
RF Random Forest
Kenya Market
  • Python
  • scikit-learn
  • Pandas
  • Random Forest
  • Feature Engineering
Blog Post Source Code
ROC Curve
AUC = 0.842 0.0 1.0
Featured Project

Credit Intelligence

End-to-end credit risk scoring system using LightGBM to predict loan default probability. Features a live Streamlit app with SHAP explainability, risk band classification, and loan recommendations — trained on 307,511 samples.

0.75 ROC-AUC
307K Samples
Live Streamlit App
  • Python
  • LightGBM
  • SHAP
  • Streamlit
  • scikit-learn
Blog Post Live App
Risk Distribution
Low Med High Very Risk Band Classification
03

Skills & Certifications

🐍

Python & Machine Learning

Proficient in Python with hands-on experience in ML algorithms, model building, and data manipulation using scikit-learn, TensorFlow, and Pandas.

View Certification
📊

Data Visualization

Skilled in creating impactful visual representations using Matplotlib, Seaborn, Plotly, and Tableau to communicate insights and support data-driven decisions.

🗄️

SQL & Databases

Experienced in writing complex SQL queries, database design, and data extraction from relational databases with practical experience in data manipulation and analysis.

View Certification
🎙️

Data Storytelling

Ability to translate complex data insights into compelling narratives. Experienced in presenting findings through clear visualizations and actionable recommendations for stakeholders.

Tools & Technologies

Python scikit-learn TensorFlow LightGBM SHAP Pandas NumPy Matplotlib Seaborn Plotly Tableau SQL Streamlit Jupyter Git
04

Get in touch

What's next?

Let's build something together.

I'm open to data science and machine learning roles, collaborations, and projects. Whether you have a position, a problem worth solving, or just want to connect — my inbox is always open.

Say Hello
LinkedIn GitHub +254 718 647 283

Designed & Built by Benedict Bett

benedictbett08@gmail.com