Interactive experiments
Machine learning, running in the browser.
A growing collection of live experiments: object detection, image segmentation, language tooling, and prediction—built so the work can be inspected, not just described.
Bello Mayowa · Applied AI & software engineer
Software that learns
to meet the world.
I build product-grade software with Python, React, Next.js, and React Native—and develop machine learning systems for visual perception, robotics control, and reinforcement learning.
A system in motion
•
observe
learn
move
From responsive products and distributed systems to the intelligence inside physical machines.
In the lab
The work, in motion.
Experiments are documented as they develop: the environment, the policy, what worked, and the limitation that defines the next iteration. Every card links to the original LinkedIn demo.
CARLA · reinforcement learning · 3:39
Highway route-following
A continuous-control agent follows a waypoint route from ego-frame geometry and vehicle dynamics. The reward balances progress, speed, steering, stalling, and lane position.Watch on LinkedIn
Unitree G1 · SAC · 1:07
Humanoid locomotion
Soft Actor-Critic learns torque commands for the G1’s 23 actuated joints. The demo documents the role reward shaping played in reaching more stable forward motion.Watch on LinkedIn
Computer vision · CARLA · 2:17
Depth, made inspectable
A look inside a depth-estimation pipeline: RGB input, predicted depth, feature maps, and reconstruction—built to make spatial reasoning easier to examine.Watch on LinkedIn
Perception · MiDaS ideas · 6:04
Dense depth in CARLA
A longer exploration of dense depth prediction from a 2D camera feed in a simulated driving environment.Watch on LinkedIn
Complete video archive11 more LinkedIn demos
Computer vision · 0:19Edges and corners Reinforcement learning · 1:21Soft Actor-Critic study Robotics control · 1:03TD3 quadruped control Self-supervised learning · 1:42Variational autoencoders Reinforcement learning · 1:18DDPG continuous control Machine learning · 1:23Sequence-to-sequence attention Computer vision · 0:59Portrait segmentation Language model · 0:30Browser email autocomplete Reinforcement learning · 0:49Game-playing RL: part three Reinforcement learning · 1:18Game-playing RL: part two Reinforcement learning · 1:04Game-playing RL: part one
Selected work
Build the foundation.
Follow the question.
Open the playground
Production software
Systems people rely on.
Seven years building full-stack products with Python, React, React Native, real-time infrastructure, and the operational details that make software hold up.
Where I'm headed
The next chapter is embodied.
I'm not trying to leave software behind. I'm bringing product instincts and systems depth to the problems where models must understand and affect the physical world.
Production foundation
Python · Django · React · React Native · TypeScript · PostgreSQL
Now exploring
Computer vision · robotics control · reinforcement learning
How I work
Prototype the model, build the system around it, then make the interaction clear.
Experience
A record of shipping.
10 Hour Labs
Senior Software Engineer
2020 — nowGitstart
Full-Stack Developer
2020Touchcore Technology
Backend Developer & Team Lead
2019 — 2020Anyskillz
Backend Developer
2019Let's talk
Have a useful problem
to work on?
I'm open to full-stack roles, applied ML opportunities, and collaborations that move closer to capable, real-world robotics.
bmayowa25@gmail.com