Machine Learning Engineer focused on building real-world AI systems across Computer Vision, NLP, Deep Learning, and Cybersecurity. My projects span satellite-based wildfire detection using YOLOv8, IoT network intrusion detection, real-time emotion recognition, stock price forecasting with LSTM, and full-stack web applications. I enjoy solving problems that matter — turning raw data into working, deployable systems.
- 🔥 Built a Wildfire Detection System using YOLOv8 + Sentinel-2 satellite imagery
- 🛡️ Built a Network Intrusion Detector — 25M+ IoT connections, 12 malware captures, ~99% ROC-AUC
- 📈 Built a Stock Price Predictor — LSTM with 5,884 tickers + 15 cryptos, live via yfinance
- 🎭 Built a Real-Time Emotion Detector — 7 emotions live from webcam (CNN + OpenCV)
- ⚽ Built a Football Sentiment Analyser — 214,738 records, 89.9% accuracy (Random Forest)
- 🏥 Built a Football Injury Risk & Performance AI — calibrated GBM (AUC 0.88) + Streamlit dashboard
- 📫 Reach me at ovoweroitoje@gmail.com
- ⚡ Fun fact: I build ML systems and watch football
|
YOLOv8 object detection on Sentinel-2 satellite imagery. Validated on real wildfire events (Rhodes 2023, Yellowknife 2023). dNBR severity mapping + Streamlit dashboard. YOLOv8 · OpenEO · Rasterio · Streamlit · Python |
IoT traffic classification across 12 CTU-IoT-Malware captures (25M+ connections). Detects C&C, DDoS, PortScan & Attack traffic. Random Forest with top-25 feature selection. Scikit-Learn · Flask · IoT-23 · Feature Engineering |
Multi-feature LSTM with live yfinance data. 5,884 stock tickers + 15 cryptocurrencies. Monte Carlo Dropout uncertainty bands, RSI, moving averages, volume chart. TensorFlow · LSTM · yfinance · Flask · Chart.js |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🔥 Wildfire Detection | YOLOv8 on Sentinel-2 satellite imagery — real wildfire detection + dNBR severity mapping | YOLOv8 · OpenEO · Rasterio · Streamlit | View |
| 🎭 Emotion Detection | Real-time 7-class emotion recognition from webcam — CNN trained on FER2013 (35,000+ images) | TensorFlow · Keras · OpenCV · FER2013 | View |
| 🌿 Crop Disease Detector | EfficientNetV2 classifies 15 plant diseases from leaf photos — upload & get treatment advice | EfficientNetV2 · TensorFlow · Flask | View |
| 🔬 Skin Disease Classifier | EfficientNetB0 transfer learning on ISIC dataset — Benign vs Malignant skin lesions, 83.1% AUC | EfficientNetB0 · TensorFlow · Streamlit · ISIC | View |
| 🖼️ Image Classifier | Upload any photo and get predictions using EfficientNetV2 deep learning model | EfficientNetV2 · TensorFlow · Flask | View |
| 👤 Face Recognition Login | Passwordless webcam login — OpenCV LBPH face recogniser, register & authenticate in real time | OpenCV · LBPH · Haar Cascade · Flask | View |
| 🩻 XRayVision | Real-time AR X-Ray scanner using MediaPipe Holistic — 4 scan modes, 10K particles, video recording. Single HTML file, zero dependencies | MediaPipe · WebGL · JavaScript · HTML5 | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🛡️ Network Intrusion Detector | Classifies IoT traffic as Benign or Malicious (C&C, DDoS, PortScan, Attack) — 12 CTU-IoT-Malware captures, 25M+ rows, ~99% ROC-AUC | Random Forest · Scikit-Learn · Flask · IoT-23 | View |
| 👁️ Fraud Eye | Advanced AML fraud detection — heuristic rules, real-time alerts, network graph visualisation across React + Node.js + MongoDB stack | React · Node.js · MongoDB · Express · D3 | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| ⚽ Football Sentiment Analysis | 4 NLP models on 214,738 Reddit + ESPN records for 15 Premier League players — Random Forest 89.9% accuracy | TF-IDF · scikit-learn · VADER · PRAW · Flask | View |
| 📰 Fake News Detector | 4 NLP classifiers detect real vs fake news using TF-IDF with bigrams — Logistic Regression 99% accuracy | TF-IDF · LR · PassiveAggressive · LinearSVC · Flask | View |
| 📧 Spam Email Detector | 4 NLP models detect spam vs ham — F1 Score metric (not accuracy) for imbalanced 13% spam rate | TF-IDF · NaiveBayes · LR · LinearSVC · Flask | View |
| 🤖 Chatbot Web | Neural network chatbot rebuilt as a Flask web app with real-time chat bubble UI | Neural Network · Flask · Python | View |
| 🗣️ RASA Chatbot | Conversational AI chatbot built with the RASA framework | RASA · Python · NLU | View |
| 🎤 NLP Personality Analysis | Comparative speech analysis of Barack Obama vs Hillary Clinton — sentiment, emotion, POS tagging, NER and personality insights | NLTK · spaCy · VADER · Jupyter · Python | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🎙️ Speech Emotion Recognition | Detects 8 emotions from voice audio — MLP trained on RAVDESS (1,440 files, 24 actors), MFCC + Chroma + Mel features | librosa · MLP · Scikit-Learn · Flask · RAVDESS | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 📈 Stock Price Predictor | Multi-feature LSTM forecasts 5,884 stock tickers + 15 cryptos with live data, uncertainty bands, and volume chart | TensorFlow · LSTM · yfinance · Monte Carlo · Flask | View |
| 🚆 Voyage — Ticket Price Forecasting | Containerised railway price prediction stack — 5 ML models, FastAPI backend, Streamlit UI, MLflow tracking, fully Dockerised | FastAPI · Streamlit · MLflow · Docker · XGBoost | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🌊 Flood Detection | Sentinel-1 SAR flood mapping — Lee filter + log-ratio change detection + Otsu thresholding across 5 real flood events with km² area calculation | SAR · Sentinel-1 · OpenEO · Streamlit · Python | View |
| 🌪️ Disaster Detection | Real-data disaster detection — flood analysis via Sentinel-2 satellite imagery and earthquake detection via the USGS API | Sentinel-2 · USGS API · GeoPandas · Python | View |
| 🏚️ FEMA Predictions | ML system predicting federal disaster recovery costs using FEMA OpenFEMA public data | Scikit-Learn · OpenFEMA · Pandas · Python | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 💉 Diabetes Classifier | 7-model comparison (LR, RF, SVM, KNN, GBM, XGBoost + Tuned RF) on Pima Indians dataset — GridSearchCV tuning | XGBoost · Scikit-Learn · GridSearchCV · Flask | View |
| ❤️ Heart Attack Predictor | End-to-end classification pipeline on UCI heart disease data — Stratified 5-Fold CV, ROC-AUC evaluation | Scikit-Learn · StratifiedKFold · ROC-AUC · Flask | View |
| 🫀 Heart Disease Predictor | Classification model predicting the presence of heart disease from clinical features | Scikit-Learn · Flask · Python | View |
| 🎗️ Breast Cancer Analysis | Full analysis pipeline on 569 patients — 97.5% accuracy, clustering, PCA, ROC curves, full documentation | Scikit-Learn · PCA · KMeans · Streamlit · Python | View |
| 💳 Loan Default Predictor | XGBoost vs 3 models on credit risk data — ROC-AUC metric, 5-fold CV, feature importance analysis | XGBoost · Random Forest · GBM · Flask | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🏥 Football Injury Risk & Performance AI | End-to-end ML system predicting injury risk (calibrated GBM, AUC 0.884) and expected match rating (regression) — synthetic dataset, time-aware splits, Streamlit dashboard for sports-science staff | Scikit-Learn · GBM · Calibration · Streamlit · Matplotlib | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| 🏥 Medical Cost Predictor | 5 regression models predict healthcare insurance costs — smoker vs non-smoker impact analysis | Linear · Ridge · Lasso · RF · GBM · Flask | View |
| 🏠 House Price Predictor | Regression model predicting house prices from location, size, and feature data | Scikit-Learn · Flask · Python | View |
| 🚗 Car Price Predictor | Predicts used car market value from mileage, brand, year, and condition features | Scikit-Learn · Flask · Python | View |
| 📦 Custom Dataset Predictor | Upload any CSV dataset — auto-detects target column and trains a prediction model on the fly | Scikit-Learn · Pandas · Flask · Python | View |
| Project | Description | Tech | Repo |
|---|---|---|---|
| Full-stack travel booking app — browse destinations, create plans, book trips, Stripe payments, reviews | React · Express · MongoDB · Stripe · Tailwind · JWT | View |