HN Explorer

Ask HN: Career Advice Needed

2 points by ta473589 ↗ HN
Hi everyone. I don't normally ask career advice of people I don't know, but I respect this community, so here goes.

I'm a software engineer with about 20 years of experience. I've worked in large and small orgs; on a website/service with 100-500 million monthly active users, and another with 10-50m MAUs; and on Big Science. I work mainly in distributed systems, and have branched out to database implementation in the past 5 years. I really love distsys.

The main economic forces that I currently see operating on my career are: (a) the ascendance of the small handful of cloud computing providers; and (b) the remaking of many facets of the economy by AI, much of this to take place in the coming decades.

On (a): I'm happy in my current job, but have found that only companies at the largest scale want to pursue the distsys approaches that I'm most interested in (e.g., gossip/CRDTs, consensus algorithms, etc.). Outside of those companies, it's difficult to get funded for anything besides, for example, leader election, or adopting a 3rd party distributed data store or cache.

OK, so on to my question... Should I:

(1) Stay where I am. (I'm lucky to work on something that I enjoy, even if it's not an exact match.) (In other words: don't worry about your 1st world problems.)

(2) Try to find distsys work at one of those large cloud computing providers.

(3) Branch out into another area, e.g., distributed machine learning, or data/compute management/scheduling systems for AI.

(4) Something else I haven't thought of.

What do you think?

I'm stuck in thinking about what to work on. Thanks in advance for your help!

7 comments

[ 3.1 ms ] story [ 21.6 ms ] thread
autodidacticon
but have found that only companies at the largest scale want to pursue the distsys approaches that I'm most interested in (e.g., gossip/CRDTs, consensus algorithms, etc.).

Can this be disproven? Are there companies of smaller scale that are attempting to solve these problems?

ta473589
There are definitely companies at smaller scale that work on these problems. I would say that the only time it really feels solved to me is if the scale is very large.
ta473589
I would also say, having worked a company of smaller scale trying to solve distsys problems, that they're hard to sell. But the distsys skill set does seem (to me) marketable at the large cloud providers.
sokoloff
If you’re happy at your current job and I don’t read anything else in your post that’s compelling you to change, why change? There are plenty of not great jobs and if you’re happy now...
ta473589
I had to leave some detail out because of the 2k character limit, which is that though I'm happy at my current job, I also feel somewhat at loose ends.

Your point about there being plenty of not great jobs is well taken. As you can see from option (1) in my list, I often lean that way myself.

PaulHoule
Some technology has trouble scaling coming and going. Like FPGA. If you are going to make something with FPGA but will sell more than 10,000 units you will be dramatically cheaper with ASIC. Each application that is is developed on FPGA will sell a small number of units so multiply that by a small number of applications and that is a small number.

Similarly there is not a market for that many large distributed systems.

What there is an endless market for is new applications from the medium to small side. For instance there is the resurgence of "low code", "no code", and many participants who feel underserved with current technology compared to what Google gets.

See this guy

https://www.servethehome.com/

as a transitional figure between "modern data centers" and "buy any 1 liter PC off ebay and see what happens".

The Baltimore Police Department keeps track of incident reports in a big Lotus Notes database, so do many 3-letter and more-letter agencies in the area south of there. Lotus Notes had some algorithms for merging changes that people make to "objects" with a very JSON-like structure that meant people built amazing things in the 1990s. (e.g. 10 years ahead of Salesforce.com!)

It must have been patents that kept the technology out of other products but it seems those must be expired by now.

ta473589
Thanks for the insight. I have often wondered if there is a space between the companies operating very large data centers and those running small businesses. In distributed systems, my own experience is:

- the very large organizations will engineer whatever they must to meet their availability/performance targets - the mid-size orgs will use whatever they can from the big orgs to work in their area of competence - the small orgs are happy with the basics

There may well be a middle ground that I should explore more.

Thanks for the servethehome.com reference. I'll try to learn from his approach.