Instrumental
paid
AI inspection and failure analysis for electronics assembly. Cameras capture every unit at each build state, then models flag defects no golden-sample rule ever encoded.
manufacturing computer vision data analysis
+1
Start anywhere
Four ways in
Building
Built, shipped,
and documented.
Systems taken from problem to production and written up honestly: approach, failures, outcome. Logs are published after a system ships, never before.
The work spans the full stack of making intelligence useful: models and agents, the software around them and the hardware they end up driving.
The anatomy of a build log
Four parts. Always the same four.
- 01
The problem
What the system exists to do and why the obvious approach was not enough. Constraints first.
- 02
The approach
Architecture and key decisions: what was chosen, what was rejected and the reasoning for both.
- 03
What broke
The failures on the way, documented as phenomena: what went wrong, why it happened, how it was detected.
- 04
What shipped
The working system, photographed and measured. Real hardware and real terminals, never renders.
Instruments
AI tools, tested on real work
Cursor
freemium
A VS Code fork built around agentic editing, with its own Composer models, a pull request review bot and SOC 2 controls for teams that want AI in the editor.
code generation software development productivity
Roboflow
freemium
End-to-end computer vision platform for labelling, training and deploying detection models to cloud or edge hardware like Jetson, with a visual builder for multi-step vision pipelines.
computer vision manufacturing automation
+1
Work
Projects, grouped by what they are
Applied work, community projects and open source get the same write-up: the problem, what was tried, what it cost. Where there is a repository, the code is linked.
Applied practice
3 projectsVentures and applied work, written up for the transferable lesson rather than the specifics.
Shenzhen, China
Presenting technical work in a second language changes the preparation
Second place at Tencent's OpenClaw hackathon in Shenzhen, pitched and defended in Mandarin. What changes about preparation, and what reading Chinese sources first-hand actually buys.
Shenzhen, China
The China Africa Trade Platform I Built That Nobody Paid For
I spent roughly two years building a China Africa trade platform and never sent a single invoice. The corridor problem was real. My method for testing it was not.
Your sourcing agent is paid to make the order bigger
Sourcing agents are paid a percentage of what you spend, so every good procurement decision costs them money. What engineer-led, fee-based buyer representation actually involves instead.
Community and impact
4 projectsWork with no invoice attached, documented to the same standard as everything else.
Mentoring at the Global Youth AI Hackathon
Mentored young builders at the Tencent-organised Global Youth AI Hackathon in Shenzhen, working on scoping, harness design and the difference between a demo and a system.
Shenzhen and Hong Kong
How to Start a Tech Community in a City Where You Are Foreign
The mechanics nobody writes down: why first meetings fail as coordination problems, why narrow definitions travel further than broad ones and what the Greater Bay Area gives an African engineer.
Shenzhen, China
Volunteer Organisation Succession Planning Starts With a Membership Record
Volunteer bodies decay when the rules live in one person's head. The order that prevents it: membership record and constitution first, programmes last, with a gate between every step.
Lagos
Teaching Scratch Programming to Secondary School Kids in Nigeria
A volunteer-led initiative that introduced Scratch programming to secondary school students in Lagos, Nigeria, empowering young minds with foundational coding skills and building a generation of future-ready digital creators."
Build logs
The first logs are in write-up.
A build log lands here after its system ships, with photographs of the rig and an honest account of what broke. The projects above are published to the same standard; the logs add the hardware.
Bench & stack
What the work runs on
The working stack behind these systems, plus the one unfair advantage: a hardware supply chain an hour away in any direction.
Languages
- Python
- Go
- TypeScript
- JavaScript
- C#
- Java
AI systems
- LLM agents & tool use
- RAG pipelines
- Machine learning systems
- Model evaluation
Platforms
- Enterprise SaaS
- WeChat mini-apps
- DevOps & CI/CD
- Manufacturing integration
Physical
- Cobot actuation
- IoT & sensor loops
- Industry 4.0 systems
- Shenzhen hardware supply chain
Thinking through a system
like these?
I'm open to a conversation about the subject matter.
Start a conversation