Loop engineering sounds great, but it also sounds like a sales pitch to burn more tokens. I am for meshing token expensive models with open models to be cost efficient, but then this becomes model herding and extra cognitive load. I remain a fan of spec driven development.
@apagajewski That robot that rocked forward and backward then asked for help piqued my curiosity. These small details shows what a small team backed by their own hardware and software can do!
I've seen really smart code reviewers. I think they'd be the best vibe coders too. Which beings me to point, I think experienced engineers are the ones best positioned for the extra AI capabilities because they know what to expect.
"AI will replace mid-level engineers". Then there will be no more senior engineers since nobody will pass mid-level. Thought leaders and their thoughts.
Apple Silicon is like AWS Nitro cards in one chip. Perhaps AWS' next step is to create their own chip for faster on-chip communication instead of going through PCIe. Also helps reduce power consumption.
Hi, I know 45bajillion queries per second is nice. Can you also tell me, did you have to scale your resources? By how much? What's your transaction latency distribution compared to regular season?