Professional Summary
Results-driven ML & Generative AI Engineer with strong expertise in predictive analytics, AI automation, and data-driven decision-making. Proven ability to design and deploy scalable AI solutions across healthcare, education, and digital automation. Skilled at LLM fine-tuning, RAG system development, end-to-end workflow automation, and building intelligent systems for real-world impact.
Work Experience
- Built and deployed an intelligent student chatbox (chatbot · widget · dashboard) serving distance learners at Ahmadu Bello University and Babcock University
- Developed the Edutech AI Assistant n8n workflow powering real-time student support and content delivery
- Engineered a Daily AI News → LinkedIn Auto-Poster pipeline that automatically sources, curates, and publishes AI news to LinkedIn
- Automated email drafting, form responses, and social media publishing using generative AI
- Integrated Google Sheets automation workflows for data capture and operational reporting
- Developed AnaRadiologyAI, a TypeScript-based AI chatbot specialising in radiology Q&A
- Fine-tuned Mistral 7B LLM on radiology-specific data, achieving +22% diagnostic accuracy and −35% hallucination rate
- Built a Pneumonia Detector & First Aid Advisor using chest X-ray imaging with preliminary diagnostic reports
- Designed a Multilingual Emotion Action Analyzer for cross-language sentiment and intent detection
- Fine-tuned Phi-4 LLM on the cosmopedia dataset for domain-specific language tasks
- Managed 10,000+ data entries, improving reporting efficiency and accuracy
- Conducted competitor analysis and boosted market positioning by 20%
- Streamlined workflows using Excel and Python
- Collected, cleaned, and analysed scientific data for government reporting
- Built dashboards in Power BI and Excel for key performance metrics
Projects
AI-powered distance learning chatbot serving students at Ahmadu Bello University & Babcock University. Deployed live on Vercel with real-time conversational support — includes embeddable widget and admin dashboard.
LangChain-powered RAG system enabling conversational Q&A over multiple local PDF documents through semantic knowledge retrieval. Built in Python.
Fine-tuned Mistral 7B on radiology-specific data achieving +22% diagnostic accuracy and −35% hallucination rate. Deployed for offline AI-assisted diagnostics in low-connectivity settings.
End-to-end n8n workflows automating emails, form responses, and social media publishing. Converts YouTube content into AI-generated infographics published automatically across platforms.
Medical imaging AI model analysing chest X-rays to generate preliminary diagnostic reports and first-aid recommendations.
TypeScript AI chatbot specialising in radiology Q&A. Deployed as part of the AnaburyAI Technology product suite at anaai.tech.
Constructed an LLM from first principles including tokenization, transformer architecture, training loop, and optimization.
Fine-tuned Qwen3 architecture on the TinyStories dataset with a custom tokenizer; includes training optimization and evaluation.
Phi-4 LLM fine-tuned on the cosmopedia dataset for domain-specific language generation and downstream tasks.
Emotion and intent detection system operating across multiple languages using NLP classification pipelines.
AI-powered educational tutoring application designed to support A-Level students with personalised learning assistance.
Comparative benchmarking of Gradient Boosting, Random Forest, and SVM across music analytics and weather/rainfall prediction datasets.
Applied Random Forest, XGBoost, and AdaBoost for regression prediction of wireless network path loss and atmospheric attenuation at mmWave/Sub-THz frequencies. Evaluated with RMSE, MAE, R².