Mechatronics Engineer interested in the intersection of Robotics and AI, and I also build Machine Learning systems that solve problems
Outside of building, I write The Nerd Stack, a newsletter on robotics, MLOps, and ROS2, and occasionally on Medium.
Paste a traceback, get a fix. Building GramAI, an offline Python debugging assistant, for Gate 1 (Aug 25).
Graduating with a capstone project on Cika, an autonomous ROS2-based waste-collecting robot.
Seven-month internship in Port-Harcourt covering data acquisition, demand forecasting, and Power BI reporting.
Maintenance of oil field equipment for Coiled Tubing and High-Pressure pumping services.
Built and documented a eight-container MLOps stack for real-time credit card fraud detection.
Data analytics and MLOps internships, including containerized model deployment with Docker and Kubernetes.
4WD skid-steer AMR running ROS2 Humble with SLAM Toolbox, Nav2, and EKF sensor fusion. RPLIDAR + OAK-D Lite for perception, YOLOv8 on-device waste detection, 6-DOF arm via MoveIt2. Capstone project, compute split across Raspberry Pi and laptop over a UART-to-ESP32 control bridge.
End-to-end MLOps platform for credit card fraud detection. Nine-container Docker Compose stack: Airflow, MLflow, Feast, FastAPI, PostgreSQL, MinIO, Redis, Streamlit. Sub-10ms feature serving, daily automated training pipeline.
Python traceback debugging assistant built on Qwen2.5-3B via llama.cpp, with a two-tier RAG corpus for grounded fixes. Built for the Africa Deep Tech Challenge 2026.
Comparative study of BFS, DFS, A*, and a Genetic Algorithm on static and dynamic grid environments. Custom C++ visualization engine built with Raylib, benchmarked across a 100x100 grid with moving obstacles.
A look at how VLA models bridge perception and physical movement, from RT-2 to pi0.
