David
Ogunmola.
Deep learning · Reinforcement learning · Embodied AI
I'm a computer engineer who likes making cheap hardware do expensive things: a $50 drone that flies on my code, 3D scenes from twelve photos, a tutor that helps on day one.
Rigorous enough to publish. Pragmatic enough to ship.
I'm a Computer Engineering student at Covenant University, fascinated by the space where deep learning meets the physical world: robotics, neural rendering, autonomous agents.
I value collaborative work and a research-first mindset: careful experiments, clear writing, reproducible code. Right now that means 3D Gaussian Splatting, imitation learning, and on-device inference on FPGAs.
Away from the bench, I play a lot of games, both competitive and casual.
Things I've built that matter.
See all 21 projectsA research stack with execution muscle.
Four domains, one continuum: from training loops to register-transfer on silicon.
AI & machine learning
Custom training loops, deep RL policies and 3D Gaussian Splatting pipelines, from theory to reproducible experiment.
- PyTorch
- scikit-learn
- MindSpore
- CANN
- RL
- Diffusion
- LLMs
Programming
Python for research, C and C++ for systems work, JavaScript for full-stack product work.
- Python
- C
- C++
- JavaScript
- Java
- MATLAB
Engineering tools
FPGA synthesis and HDL for on-device signal processing; hardware-in-the-loop simulation.
- Simulink
- VHDL
- HDL Coder
- FPGA
- Arduino
- Fusion 360
- AutoCAD
Web & cloud
Next.js, React Three Fiber and Flask or Node APIs: research tooling that ships.
- React
- Next.js
- Tailwind
- Node.js
- Flask
- Azure AI
- GCP
Ask the network.
A real four-layer network, small enough to read. Tell it what you care about and it runs a forward pass, layer by layer, to the project you'll probably like most. Every line is a real weight: orange pushes up, blue pushes down.
Write to me.
If you're working on robots, rendering, or anything that has to run on hardware cheaper than it should be, I'd like to hear about it.