Data Scientist

Ireoluwawolemi
Jeremiah Akindipe

ML Engineer·Python Developer·Analytics

Transforming raw, complex datasets into actionable business intelligence. Specializing in predictive modeling, time series forecasting, and statistical analysis across pharma, retail, and market domains.

View Projects LinkedIn ↗
14 repos
Projects
98 %
XGBoost R²
6 domains
ML Areas
Open
to collaborate
Ireoluwawolemi Jeremiah Akindipe
Open to opportunities
analyst_dashboard.py running
Projects
14 repos
→ active
Primary Stack
Py 3.x
+ Scikit-learn
ML Methods
10 +
classification, regression…
Status
Open
to collaborate
Skill Proficiency
Python
94%
Pandas
90%
Scikit-learn
88%
Visualization
84%
Time Series
76%
01 //

About

Building Intelligence from Data

I'm a data scientist passionate about uncovering the stories hidden inside complex datasets. My work covers the full ML pipeline from exploratory analysis and feature engineering, through model development and rigorous evaluation, to visualization and stakeholder-ready output.

I've applied these skills across pharmaceutical analytics, retail demand forecasting, customer segmentation, quality prediction, and real estate market analysis.

  • Predictive modeling regression, classification, ensemble methods
  • Time series forecasting ARIMA, Prophet, feature engineering
  • Unsupervised learning K-Means clustering, PCA, dimensionality reduction
  • Exploratory data analysis and statistical modeling
  • Interactive dashboards and data storytelling
Core Language
Python Jupyter Numpy Pandas
Machine Learning
Scikit-learn XGBoost Random Forest K-Means PCA
Visualization
Matplotlib Seaborn Plotly Streamlit
Currently Learning
TensorFlow Docker MLOps
01.5 //

Certifications

TS Academy Certified: April 01, 2026
Data Science
Professional Certificate
Successfully completed a rigorous 3-month intensive training programme in Data Science, satisfying all programme requirements across machine learning, statistical modeling, time series analysis, and data visualization.
Duration: 3 Months Intensive Programme
Issued by: TS Academy Adetola Durojaiye, Founder & CEO
Credential ID: TSAH9FR9SD6KDH
// credential id: TSAH9FR9SD6KDH
TS Academy Data Science Certificate
02 //

Projects

Production ML · Fraud ↗
E-Commerce Fraud Detection System
Production-grade end-to-end ML pipeline for detecting fraudulent transactions. Time-aware chronological splits, 17 engineered features, XGBoost primary model with Isotonic calibration, cost-sensitive thresholding (FP:5 / FN:100), and full SHAP explainability. Live FastAPI service on Render with health and docs endpoints.
XGBoost
Primary Model
SHAP
Explainability
15K
Transactions
Calibrated
Isotonic + Cost
Python XGBoost SHAP Scikit-learn Calibration Feature Eng. Joblib
LIVE API · Backend API Docs GitHub
// Project Visualizations
Production tuned XGBoost confusion matrix
Confusion Matrix · Production Threshold
SHAP beeswarm summary plot
SHAP Beeswarm Explainability
Calibrated precision-recall curve on test set
Precision-Recall Curve (AUC 0.62)
Live FastAPI service on Render
Live FastAPI on Render
Full-Stack · ML App ↗
Smart Career Predictor Portal
End-to-end career guidance platform for Nigerian secondary-school students. FastAPI backend + React/Streamlit frontends, XGBoost multi-class career classifier (85%+ test accuracy, 88% stratified CV), JWT auth, SQLite persistence, and Google Gemini narrative generation with offline fallback. Docker-ready and production-deployable.
85%+
Test Accuracy
88%
5-Fold CV
10
Career Classes
FastAPI
+ Gemini LLM
XGBoost FastAPI React Streamlit Docker Gemini JWT
LIVE APP · Frontend API · Backend GitHub
// Project Visualizations
Smart Career student dashboard
Student Dashboard & Journey Map
Career pathway recommendation report
AI Career Pathway Report
XGBoost classification report heatmap
XGBoost Classification Report
FastAPI health check response
FastAPI v2.0 Health Check
Research · Meta-Learning ↗
Meta-Learning for Algorithm Selection
Empirical meta-learning study on ~20 PMLB datasets. Extracted dataset meta-features, trained five base learners (RF, GB, SVM, k-NN, NN) under stratified 5-fold CV, ranked algorithms, and trained a meta-learner that predicts the best algorithm from meta-features alone. Full statistical validation with Friedman + Nemenyi and Wilcoxon tests plus IMRAD manuscript.
5
Base Algorithms
~20
PMLB Datasets
Friedman
+ Nemenyi
Meta-Learner
Trained
Meta-Learning Scikit-learn PMLB Statistics Research Reproducible
VIEW REPO · Empirical Study
// Project Visualizations
Nemenyi critical difference diagram
Critical Difference (Nemenyi)
Dataset complexity vs best algorithm performance
Complexity vs Performance
CV score distribution by algorithm
Score Distribution by Algorithm
Per-dataset algorithm rank heatmap
Per-Dataset Algorithm Ranks
EDA · ML · Deployment ↗
Instagram Engagement Intelligence Platform
End-to-end engagement-rate prediction system. Comprehensive EDA across media types, categories and traffic sources, 8 engineered ratio features, tuned XGBoost regressor, and a four-tab interactive Streamlit dashboard (Overview · EDA · Model · Predict) for pre-publish forecasting.
XGBoost
Tuned Regressor
8
Engineered Feats
4-Tab
Dashboard
Live
Streamlit App
XGBoost Streamlit Feature Eng. EDA Scikit-learn Power BI
LIVE DEMO · Streamlit App
// Project Visualizations
Instagram Analytics dashboard overview
Engagement Intelligence Dashboard
Actual vs predicted engagement rate scatter plot
Actual vs Predicted (XGBoost)
XGBoost time series feature importance
Time Series Feature Importance
2026 engagement rate forecast
2026 Engagement Rate Forecast
Time Series · Forecasting ↗
Walmart Weekly Sales Forecasting
Capstone forecasting engine for Walmart weekly sales across 45 stores. Hybrid approach combining tuned XGBoost (R² 0.98, MAE reduced ~59%) with SARIMAX seasonal models. Heavy feature engineering (lags, holiday proximity, markdown intensity, store-size interactions) and industry WMAE metric that weights holiday weeks 5×.
0.98
XGBoost R²
SARIMAX
Seasonal Model
45
Stores
WMAE
Holiday-Weighted
XGBoost SARIMAX Time Series Feature Eng. WMAE Retail
VIEW REPO · Capstone Forecasting
// Project Visualizations
Retuned SARIMAX forecast with exogenous variables
SARIMAX Forecast · Store 14 Dept 92
Weekly sales trend with 12-month moving average
Weekly Sales Trend (12-month MA)
XGBoost feature importance for weekly sales
XGBoost Feature Importance
Actual vs predicted weekly sales scatter plot
Actual vs Predicted Weekly Sales
03 //

GitHub Stats

📁
14 repos
Public Repositories
→ actively maintained
⚙️
100 %
Python / Jupyter
primary language stack
🧠
6 domains
ML Domains Covered
fraud · career · meta · retail
Skill Proficiency self-assessed
Python94%
Pandas / NumPy90%
Scikit-learn88%
Data Visualization85%
Time Series82%
Unsupervised ML84%
Statistical Analysis84%
Project Domain Breakdown
Predictive Modeling 35%
Time Series 15%
Full-Stack / Deploy 25%
Meta-Learning / Fraud 25%
Contribution Activity 2025-2026
180+
Total Commits
14
Repos
2025
Active Since
04 //

Contact

Let's build something with data.

Open to collaborations, research projects, freelance data science work, and full-time opportunities. If you have an interesting problem that needs data-driven thinking, reach out.

GitHub
github.com/A-I-Jeremiah
→
LinkedIn
ireoluwawolemi-akindipe
→
// Current Status
name Ireoluwawolemi J. Akindipe
role Data Scientist · ML Engineer
focus Production ML · Fraud · Forecasting
stack XGBoost · FastAPI · SHAP · Docker
learning MLOps · Deep Learning · LLM Apps
available Open to opportunities