A client recently tasked me with developing a quick to deploy and easy to maintain sensor for evaluating potential wind farm site efficacy. The opening frame was one of networking and communication challenges, the sensor’s accuracy was assumed. In reality the main challenge was correcting sensor readings in a dynamic environment. A tethered weather balloon experiencing wind is moving itself, you’re measuring from a moving platform.

I inherited a petabyte-scale network storage problem that kept failing under load. The team had been adding compute, distributing across servers and even transitioning to the cloud. Adding compute actually made it worse, the bottleneck was network latency. I distributed operations across network drives, optimizing for individual drive read and write speed limits.

My technical work doesn’t stop in my freetime. I recently built a chicken-coop in my yard and automated the doors to open at sunrise and safely close at sunset. I’ve been controlling motors for years, going back to my college publications, so I assumed the main challenge would be the mechanics of physically opening the doors. It turned out, however, this was a decision tree coordination problem. When the highest priority is bird safety, decisions have to be made confidently and movement can’t happen blindly.