Final-year Computer Science student at Babcock University
Machine Learning Engineer focused on production-ready AI systems
I design and deploy end-to-end machine learning systems, covering data processing, model development, API serving, deployment, and monitoring. My work emphasizes scalability, explainability, and real-world impact.
- Deep Learning (CNNs, LSTMs, Transformers)
- Recommender Systems (Collaborative, Content-Based, Sequential, Reinforcement Learning)
- NLP and LLMs (RAG, Fine-tuning, LangChain, Hugging Face)
- Explainable AI (SHAP)
- FastAPI, Flask, REST APIs
- Docker, MLflow, GitHub Actions
- Model serving and inference optimization
- Python, TensorFlow, PyTorch
- Pandas, NumPy, Scikit-learn
- Git, Linux, Jupyter
Hybrid Reinforcement Learning and Transformers | FastAPI | Docker | MLflow
Production-grade recommendation engine optimizing long-term user value through a multi-stage pipeline combining collaborative filtering, content-based models, sequential transformers, and a reinforcement learning decision layer.
GitHub: https://github.com/okefemi12/ecommerce-recommender-system
Live Demo: https://hybrid-rec.vercel.app/
LSTM | SHAP | LLMs | Streamlit
Real-time financial fraud detection system combining deep learning with explainable AI and automated compliance reporting for non-technical stakeholders.
GitHub: https://github.com/okefemi12/credit-scout-risk-engine
Flask | RAG | Firestore | OCR | Docker
AI-powered academic assistant supporting PDF and image uploads, secure user sessions, chat history storage, and student performance prediction.
GitHub: https://github.com/okefemi12/student-success-chatbot
API Test: https://student-success-backend.onrender.com/test
Designed a novel metric called Ball Pursuit Efficiency (BPE) using ensemble deep learning models (XGBoost, LSTM, Transformer) on NFL player tracking data.
Kaggle Notebook: https://www.kaggle.com/code/okeoluwanifemi/nfl-prediction
Video Demo: https://youtu.be/M6CLhlZFp0M
Fine-tuned Llama-3 to detect and fix common security vulnerabilities in Python code, achieving an 88 percent Bandit static analysis pass rate.
Hugging Face Model: https://huggingface.co/oke39/llama3-8b-secure-code
Hugging Face Profile: https://huggingface.co/oke39
- IBM Machine Learning Professional Certificate
https://coursera.org/share/95d5dcd25e20acf902f38222ae6a4b33
- LinkedIn: https://www.linkedin.com/in/oluwanifemi-precious-oke-135623321/
- GitHub: https://github.com/okefemi12
- Kaggle: https://www.kaggle.com/okeoluwanifemi
Building scalable machine learning systems, not just models.