The Newton-Muon Optimizer replication
A small implementation and experiment exploring how Newton-Muon changes optimization and validation behavior on Tiny Shakespeare.

AI Engineer & Researcher
Currently Building
AI-native banking and payments for Africa.
Browser task recording and replay for AI agents.
Shipped
A small implementation and experiment exploring how Newton-Muon changes optimization and validation behavior on Tiny Shakespeare.
A small experiment inspired by sparse memory finetuning, testing whether an LLM can learn new facts without forgetting what it already knows.

TEDx Lokoja 2026
May 16, 2026
A talk about the stories we tell ourselves that stop us from beginning the things we are most qualified for. Drawing on Carol Dweck's mindset research and the journeys of Jan Koum and Demis Hassabis.

Build With AI Lokoja
April 11, 2026
Walked through building production-ready AI agents using Google's Agent Development Kit (ADK) in Python. Covered agent architecture, tool use, and practical patterns for deploying autonomous systems.

Daytona Developers Club Tour '25
October 11, 2025
Explored the paradigm shift in software development from using AI as a tool to building AI-first applications. Discussed practical strategies for developers to embrace AI-native development workflows.

Supabase LW15 Meetup Lokoja
July 19, 2025
Shared the journey of building Datum Africa, a collaborative platform for African datasets, powered by Supabase. Discussed the technical architecture and the mission to democratize access to African data.

IndabaX Nigeria 2025
July 17, 2025
Presented a practical framework for transforming AI ideas into minimum viable products. Covered product strategy, technical implementation, and go-to-market approaches for AI startups.
“A small reproduction of Newton-Muon on Tiny Shakespeare, and why training loss can be a dangerously misleading signal when comparing optimizers.”

“I implemented a small continual-learning experiment to understand whether sparse memory can teach an LLM new facts without destroying what it already knows.”
Open to engineering and research collaborations.
Let's Build Together