Software-driven carpentry

In the past few weeks I did something different. I needed a break from my usual "projects" so decided to build furniture.

Problem definition

The business sponsors at home decided they needed some new furniture. As the resident client partner, solution architect, and generally the cheapest IT guy around, I was called in for the discussion. They wanted "cheap, fast and good". I started explaining about how you can't get all three but they all left the room after giving me the requirements.

Build vs buy

The inevitable step of build vs buy followed. Ultimately I was able to demonstrate how to leverage prior investment in platforms and tools (work benches, power saw, table saw, sander, etc.) to get a customized solution. There were some delicate questions about prior references and some quality issues with past work but I was able to quickly progress past that stage. We were able to agree to a "build" decision.

Prototype

The first step was to get approval for the solution. Using advanced software modeling tools on a GPU (buzzword bingo check!) I was built out the mockups and wireframe. 3D Builder on Windows 10 was the tool of choice, and availability.

It was going to be a platform couch/sofa with an open frame design.

Build phase

The dev environment was set up by emptying the garage of cars. Some user groups had issues parking the cars outside in the cold. I started with building the foundational elements. Needless to say, I used "agile" methodology with multiple sprints to the local hardware store.

In about a day or so I had a MVP ready. Initial user reviews were good and encouraging. Additional funding approval was gained and load testing was also done with successful results. The UI was obviously still not ready since the focus was on the framework build out. A key feature was the modular back-end that could be decoupled as needed.

User Interface

The final step was to finish out the UI. The UI consisted of an underlying preprocessing layer (pre-stain conditioner because I was using soft pine wood), followed by multiple layers of core UI (oil-based stains of the preferred shade). The presentation layer took 2 days to pass the sniff tests (multiple coats of lacquer finish tend to set off fumes).

And finally, just after Christmas Day the solution was delivered. There was a key specification deviation that resulted in a change request. Apparently, an incompetent developer read "75 inches" as "72 inches". Well, due to the use of advanced engineering principles in combination with dumb luck I could address that change in about 10 minutes with minimal rework.

OSHA Update: we needed only 2 band-aids and there were only a couple of superficial bruises.