Advice for New Principal Tech ICs (I.e., Notes to Myself)
15 comments
·October 25, 2025jackblemming
The subtle disdain I hear from these types of super elite principal distinguished architects about actually writing code amuses me. Of course actually writing code is far too lowly of an activity, they’re more of an “ideas guy”. Fred Brooks had a good laugh about these types decades ago.
shermantanktop
> While you should still be writing code
Did you miss this?
ninetyninenine
In short, you manage people. You tell others what to do and you think about what to do.
It's technical management. You create impact off the work of others. You tell 10x people to do 10x work and you get 10x the credit when management only is like 1x of talent and effort.
I met a principle engineer who didn't know what a database transaction was and it still fits every description on this site. We shouldn't call these people engineers anymore. They are managers that don't need to do 1 on 1s.
lumost
There needs to be roles for pure ICs. In deep tech orgs, it’s almost impossible to get work done unless you have a bench of folks with domain expertise, comfortable on your software, and in tune with and guiding product direction. Diluting the core IC responsibility in these orgs with mandatory influence/scope guidelines is counter productive.
That being said, when I was in the role of a principal engineer at a major tech company - my ultimate responsibility was to get the project to success. What that meant on any given week could have alignment, coding, guiding, or reviewing.
light_hue_1
> In short, you manage people. You tell others what to do and you think about what to do.
As a principal scientist I definitely don't manage people.
Mainly, people managers have power to tell people what to do and when. I have zero power over anyone. No one needs to listen to me. Ever. If people ignore me there's literally no one I can complain to. And no one tells me what to do. If I wasn't competent at my job I'd be sitting alone in an office staring at a wall.
The reason why anyone would ever talk to me is because I have well over a decade of experience more than every other scientist in my org (200+ people) doing very complicated things, at scale, across many areas of AI/ML, and publishing many dozens of papers in top venues. So I know what I'm doing and can find problems, shortcuts, keep them from wasting time, find creative fixes to AI/ML issues, and see connections to business problems. Basically help scientists deliver better features faster. I can find new projects that those scientists are excited about. And I can communicate with product, managers, leaders about very complex technical ideas in ways that more junior scientists cannot.
People in my org come to me because I provide value basically. Not because they have to.
> You tell 10x people to do 10x work and you get 10x the credit when management only is like 1x of talent and effort.
If only! 70% of the work I do is invisible small course corrections here and there across multiple orgs that fix things no one ever hears about but would be disasters down the road. 20% is working on projects that scale with many people. 10% is strategy for new experiments and projects.
For the vast majority of what I do other people get 100% of the credit and my name is a small footnote at best.
Maybe an example would help. Sorry that I need to be a bit vague. A large project recently was going in a particular direction. I saw two months ahead of time that this direction was going to run out of steam for a combination of science and business reasons. It would end up missing metrics. Leadership was asking for something that was broadly correct but subtly wrong. I spent a month finding a new direction that was the opposite of what everyone thought we had to do, finding the right small 1-2 person experiments to derisk the new direction, get senior people to understand it and agree to it, set up the correct relationships to make it viable, and then I had the evidence to convince project leadership and management to shift. A few days later we had a VP review and the response was that they're delighted with the direction, it's far better than they ever though, and this will be a flagship for an org with tens of thousands of employees. That review would have gone horribly without what I did because the problem I saw two months ahead of time would have surfaced and the project would be seen as dying and aimless. But the team got the credit for being on top of things as a whole, not me. As it should be.
> They are managers that don't need to do 1 on 1s.
I have more 1:1s than managers, because they can afford to just tell people what to do. I need to develop junior scientists, keep relationships up, stay on top of other orgs, business priorities, etc.
Managers fail by creating chaos around them. Principals fail by becoming irrelevant.
sandeepkd
From what little I have seen, this kind of role is tightly coupled and dependent on the their Manager. The manager has to you like as a person and some how believe that all these activities are adding value.
light_hue_1
> From what little I have seen, this kind of role is tightly coupled and dependent on the their Manager.
As you go up the chart you have more independence and are less tightly coupled to your manager. By the time you get to principal you should be largely independent. At the same time, you have much more responsibility.
That's just a practical problem. As your manager becomes more senior (director/VP) their scope also increases. They just cannot "manage" you the way someone would manage a more junior IC. Also at the principal level you aren't just bringing value to your manager, but to other parts of the org as well.
In other words, I can't ask my manager "what should I do today?". I cannot even imagine what his reaction would be if I asked that question.
> The manager has to you like as a person and some how believe that all these activities are adding value.
For what it's worth my manager is a great person. But he wouldn't for a moment believe anyone when they say they add value.
It's up to me to find ways to document and express my value. Figuring out how to do this is part of becoming a principal. So I keep notes, I record wins, I make sure that I do things that bring me visibility, that I present new ideas, I contribute to larger roadmaps at the org level, I make sure that other scientists can say good things about me, I help fix problems that other orgs have so that they report I was useful, etc.
re-thc
> People in my org come to me because I provide value basically. Not because they have to.
In most orgs sadly that's not why people talk to a principal. It's that the process requires it, the manager wanted them to and/or there's you i.e. someone else to take responsibility. Of course they won't tell you that directly.
cindyllm
[dead]
DiscourseFan
I guarantee you that the "10x" engineer would not be able to co-ordinate anything with any other team or department in the company. Of course the hierarchical structure of companies means that just having good people skills, and good management skills, means better compensation. But the most important skill in any capitalist economy is organizing labor, since capitalism is a way of organizing labor, so those who are the best at organizing and directing labor are the most well compensated since they are able to produce the most profits.
BobbyJo
The most competent engineer I've ever worked with, the only one I'd consider 5x+, was fantastic at cross team collaboration.
microflash
This is just another stereotype. I’ve seen plenty of managers struggling to form coherent communication where an engineer would express with perfect ease. The problem with communication comes when you add layers of bureaucracy between people, often under the guise that engineers can’t communicate.
Simon_O_Rourke
Yet another in the ever lengthening list of "have you seen how fantastic I am" ego stroking write ups disguised as a technical blog post.
Individual Contributor