It's tough to value the difference between a Python programmer with 5 years experience and 10 years experience.
The second is not worth 2x the first... but what's the delta?
The second is not worth 2x the first... but what's the delta?
The reason why is pretty straightforward but I don't hear it talked about much. Any human mind is pretty good at solving a problem, and two heads are better than one. But the bottleneck is the basic communication between people.
So the more contextual background (experience) one has in a problem, the faster one is able to explore the problem space and iterate before he or she begins writing the solution in the real world. That's not going to be replaced by Agile or any other workflow in the foreseeable future.
I saw this firsthand from my supervisor when I had a 1 year contract at hp in my mid 20s (he was in his 40s). I was used to dealing with C++, compilers, some scripting with PHP and Python, that sort of thing.
It took me about a day to complete a ticket like writing a method to, say, load some data and show it on the screen in C++ or HTML. Meanwhile he was swizzling data in spreadsheets, piping it around the shell and doing deep analysis on it in large single passes in a functional manner with the vector and matrix processing techniques of MATLAB. He'd go from unstructured data to a report in an hour or two. Stuff that would take me days, weeks, or not even be feasible with the techniques that I was used to.
I look back on that year as being as important to my education as all the years before that and college combined. My only regret is that I see so far now that it almost hurts to write the code where the money is. A deep reservoir of discipline and frankly stamina are needed to protect the mind from being drug down by the antipatterns surrounding us.
Even if that’s true, most companies are at most looking at the next fiscal year and VC backed funds are just looking to survive until their next funding round or to their exit.
I am actually a similar age to the OP. I am 49, not far off 54 in % terms. I have 19 years of Python experience.
I recently started a new job, my 3rd time in consumer finance.
Two months in, I already understand the business far better than the two year mark of those two previous roles. My direct productivity is at least 3x me ten years ago. My likely impact on the business is 10+ times more.
Plus I still work harder than virtually anyone else I know, regardless of age.
So what's the delta?
The low-hanging fruit in ML is absurdly low these days, the tools are very good and there is loads of sample code available. If you were already good with Python or R, and already had some nice clean data, you could be doing something useful with Keras in a day, no joke.
For instance, you would probably say I am only half as productive as the data scientists I manage. I only get to spend an hour or two a week doing hands-on modelling, so I just don't have the same familiarity with the techniques and tools.
But in terms of influencing output, I can have a five minute chat about the goals and context to their work and save them a week of fruitless effort. So in practical terms, my productivity is far higher.
I could easily spend a bit of time to improve my modelling skills a bit. But the law of comparative advantage suggests our junior engineers should build their skills while I coach, mentor and set strategic direction.
In fact, to give another very practical example, this weekend my company Oakam is hosting the European Pandas core developers (https://python-sprints.github.io/europandas2019/) in London. My own personal contribution to Pandas this weekend will be minor. But I am far more effective by being able to provide the venue and fund travel grants so 30 core developers can collaborate together.
> you’re probably not as good as you think you are
Here is one reference point for you... Check out the 1992 winners of the ACM Programming Competition here. I'm second from the right in the photo.