4 ms·
> surely the folks who are pulling those kinds of salaries are working on really interesting problems > ... > The really hard problems that my org faces arent t
by I_AM_A_SMURF 5y ago
> surely the folks who are pulling those kinds of salaries are working on really interesting problems
> ...
> The really hard problems that my org faces arent the same league of hard problems that someone at a FAANG company (or similar) is working on
You'd be surprised. What I work on is relatively mundane. I worked at multiple FAANG companies in the past and the work there is mostly mundane too.
For the most part the difference is in the order of magnitude of users that you have, so you get to look at edge cases more often, need to be more careful about having a stable codebase etc, but writing a Web service that gets X in input and spits out Y in output is not rocket science.
I think the carpenter/woodworker analogy for software engineering works pretty well. The very expensive carpenters do the same work as the inexpensive ones, the difference is in the details.
- theshadowknows 5y agoIf you don’t mind my asking, what would you say you spend most of your time doing in a week? For me it’s a lot of requirements gathering and sketching things out then developers go off and build it. For example next week I’ll be working with a few other architects on some identity resolution projects. But we’ll just design the system, the engineers will take our specs and go off and build it.
- I_AM_A_SMURF 5y agoI luckily still get to write code (not that common at my level), mostly gnarly undocumented bugs where you need to read the linux source code because the behavior is not documented anywhere else. But most of my time is usually split halfway between coaching and helping our junior and mid level engineers, pair programming, telling them that yes, they need all those tests, stuff like that and coordinating with other senior-er engineers on big projects, making sure we don't step on each other toes and lately optimizing the overall architecture of the product which has a lot of overhead right now (we build a big monolithic piece of software, not the easiest ship to steer).