Hi, i’m joseph
an Oshawa-based developer building browser apps, 3D simulations and on-device AI tools

















































About me
I'm Joseph Bepo, Deji to most people, based in Oshawa, Ontario. I studied Mechanical Engineering at Obafemi Awolowo University and Computer Programming at Durham College, where I made the Honor Roll. I build browser apps, 3D simulations and AI tools that run on your own device, and I bring the same care for accurate data from hands-on SAP and WMS-driven warehouse work. Let's build something useful together!
bepojoseph9@gmail.com · github.com/josephbepocode
Currently Shipping/Receiving Team Lead at Lactalis Canada, Oshawa (since June 2025).
Services
Web Apps
Fast, static-first web apps and sites in HTML, CSS, JavaScript and React, deployed on GitHub Pages, from booking sites to dashboards and storefronts.
3D & Simulation
Interactive three.js scenes that model real systems, like warehouse robots planning routes with A* and city traffic with car-following physics and signal strategies.
On-Device AI
AI tools that run in the browser with no server or API keys, using WebLLM, transformers.js, Pyodide, MediaPipe and Magenta.
Data & Visualization
Turning open data into clear tables, charts and shot maps with D3.js, plus Python analysis with pandas.
Back-End & Automation
Node.js and Express servers, Python and Flask tools, and scripts in Python, Bash and PowerShell to automate everyday work.
Project
Personal · AI Tool
Agent Forge
A multi-agent app builder that runs in your browser: a planner writes a spec, a coder writes a single-file HTML app and a reviewer fixes bugs, all on a local LLM through WebLLM.


Personal · 3D Simulation
Warehouse Digital Twin
A three.js digital twin of a warehouse floor where autonomous robots take orders, plan routes with A*, pick from racks and deliver to packing stations without colliding. All data is simulated.


Personal · 3D Simulation
City Traffic Sim
A 3D agent-based city traffic simulation where every car follows the Intelligent Driver Model, comparing fixed-time, max-pressure and green-wave signals with live KPIs and a built-in benchmark.


