Omar pfp
Omar
@omar
What would happen if we compensated infrastructure engineers like we paid a top trader, or a pro golfer? Purely based on performance and results: My company's CI pipeline is frustratingly slow. `git push origin master` to production rollout can routinely take 10 minutes. We deploy a lot, and that 10 min is a huge tax on productivity and velocity. What i'm "supposed" to do today: * Put up a job posting for an infra engineer * Sift through 100 profiles, interview 10, negotiate, close, start date, etc etc etc and * maybe in a few weeks/months, I found the right person and my CI pipeline is faster. What I want * Let's say mean build time is 600 seconds, and * Mean time to failed build is 480 seconds. I'll gladly pay someone $10 for every second of mean build time they can reduce, and $20 for every second of mean failed build time. So let's say you can get it down MBT=60s and MFBT=30s I'd gladly pay a talented contractor $14,400 to deliver those results ((600-60)*$10 + (480-30)*20). Probably more.
1 reply
42 recasts
45 reactions
Omar pfp
Omar
@omar
Sure, lots of caveats here. But let's not talk about this idea in abstract; let's talk about real-world needs. * This wouldn't need to be a FT employment agreement * And this probably wouldn't be a substitute for eventually bringing on a FT infra engineer. Continuity, roadmap, etc etc caveats, yes I know, it optimizes for short term results. but right now I am strategically thinking short term. But like even if it only takes an hour of work, all the better.
1 reply
0 recast
0 reaction
Omar pfp
Omar
@omar
or here's another example: I recently cut a ton of over-provisioned cloud resources: https://x.com/omarish/status/1828272808717111308 there was zero service degradation from this change. the only metric that went down was my burn rate. I'd gladly pay some function of LIBOR + the amount of monthly savings someone could deliver, without degrading performance.
1 reply
0 recast
0 reaction
Omar pfp
Omar
@omar
that sweet feeling when you finally turn on vertical pod autoscaling
0 reply
0 recast
0 reaction