TOCHUKWU AROH

Data Scientist & ML Engineer building production-grade machine learning systems — from raw data ingestion to deployed APIs, monitored in real time.

Python · R · SQL · PostgreSQL · Docker · FastAPI · MLflow · Shiny

Healthcare · Clinical ML · MLOps

Heart Failure Clinical Decision Support System

Production ML system predicting patient mortality risk for hospital risk teams, clinicians, and insurance assessors. Three models trained and compared — Logistic Regression selected for deployment with 89.5% recall and 0.882 ROC-AUC. Decision threshold set to 0.3 (clinically conservative) to minimize missed deaths. Returns risk level, clinical flags, and actionable recommendations per patient.

Python · XGBoost · Logistic Regression · MLflow · FastAPI · Docker · PostgreSQL (3 schemas) · Evidently AI · GitHub Actions CI/CD

Heart Failure Clinical Decision Support
NLP · Booking ML · Dashboard

Airline Customer Intelligence & Booking Optimization Platform

British Airways Intelligence Platform

Dual-module platform analyzing 1,920 customer reviews (NLP) and predicting booking completion from 50,000 booking records (ML). Topic modeling with LDA revealed delays generate 100% negative sentiment. Live KPI dashboard with 3 interactive tabs. 85.4% recall on booking prediction. FastAPI + Docker deployed.

Python · spaCy · VADER · TextBlob · LDA · XGBoost · MLflow · FastAPI · Plotly Dash · Docker · PostgreSQL

Telecom · MLOps · Production API

Customer Churn Prediction System

Customer Churn Prediction

End-to-end churn prediction system for telecom customers. Full ETL pipeline loading 7,043 records into PostgreSQL, XGBoost model with 0.84 ROC-AUC, experiment tracking in MLflow, live FastAPI prediction service, drift monitoring with Evidently AI, and automated CI/CD with GitHub Actions.

Python · XGBoost · MLflow · FastAPI · Docker · PostgreSQL · Evidently AI · GitHub Actions

R · Behavioral Analytics · Live Deployment

Mobile Device Usage & User Behavior Analysis

Mobile Behavior Intelligence

R-based behavioral classification system categorizing 700 mobile users into 5 behavioral classes. Three models trained (Logistic Regression, SVM, Random Forest) with 95.7% accuracy. Deployed as an interactive Shiny web application — publicly accessible with no setup required.

R · randomForest · SVM · caret · ggplot2 · Shiny · shinydashboard · shinyapps.io

About Me

I'm a Data Scientist and ML Engineer based in Abuja, Nigeria. I build end-to-end production machine learning systems — not just notebooks. Every project I ship has a database, deployed API, experiment tracking, drift monitoring, and automated CI/CD.

I work across Python and R, with expertise in supervised learning, NLP, clinical decision support, and business intelligence. I'm the founder of a data intelligence consulting practice serving SMEs, and I'm open to remote data science and ML engineering roles.

Core Skills: Python · R · SQL · PostgreSQL · XGBoost · Random Forest · SVM · Logistic Regression · spaCy · VADER · LDA · MLflow · FastAPI · Docker · Shiny · Plotly Dash · Evidently AI · GitHub Actions · Pandas · scikit-learn · caret · ggplot2