Most developers understand that the job interview questions they're asked right now might have little to no bearing on the actual work they'll be doing tomorrow.
The challenge is assessing how their role will change over time, and begin skilling up accordingly. The rise of data science in almost every industry offers a good example of how developers can be caught unprepared, and what they should do to chart the right course in their careers.
According to a research study conducted by a team at Google and the University of San Diego, for instance, there are often challenges for developers who initially specialized in Java to make the move into a data science role.
These included what the researchers described as a difficulty in grasping the foundations of machine learning (ML) concepts, and a feeling that they “never get a full understanding of the algorithms.”
In other cases, "(Respondents were) too often faced with mathematical equations (e.g. backward propagation), that are very hard to implement correctly," the researchers said.
While there are tools and programming workflows such as Tensorflow's Tensorboard visualization toolkit that can help here, Google researchers have since followed up with a blog post that suggests these need to embed implementation tips and just-in-time strategic pointers to the underlying math.
Of course, this doesn't help Java developers or other programming professionals who may be working for (or applying to work for) organizations that will clearly need to make use of data science over the long term. With that in mind, here are three strategies to consider that are feasible for any developer to tackle immediately:
1. Explore and experiment with ML-specific languages
Less than two weeks ago, a group of developers posted the release notes for Julia 1.4, a programming language that has been enhanced with improved multi-threading, new library features, and tweaks to the build system.
Occasionally compared with C in the speed with which it can be used, Julia has already been downloaded more than 13 million times, even if it has fewer packages than what you would learn in Python courses. The earlier developers get exposed to it, the more likely they'll be ready to answer future employer's questions.
2. Get to know your data science team mates
While developers may take the lead on ML projects and other initiatives, it's more likely they'll become part of a group that includes coworkers from a variety of academic and professional backgrounds.
Recruiting firm Dice has published insights into the day-to-day life of data scientists, whether they emerged as developers or had more of a managerial role. The one common theme is that anyone working in this area will need to be in a constant state of learning.
3. Kick-start your data science skill development
As the needs and applications for data science continue to grow, some organizations will no doubt support professional development and training activities to ensure they have the right talent in place.
In the meantime, however, developers will obviously have an edge if they seek out and acquire some of those skills on their own. There are lots of machine learning courses to choose from today, for example, that give working professionals options to study part-time, or full-time if you're not yet employed anywhere.
It may seem premature to dive deep into data science right now, but don't forget that people once thought the same thing about programming itself. Just as learning to code can help those in all kinds of roles, data science skills will become table stakes before you know it.
Top comments (3)
Yes, you're right -- I meant Tensorflow! Thanks for letting me know. I just changed it. And thanks for reading and for your feedback. Really appreciate it.
Thanks again. I tried to clarify and linked to it directly. Hope that helps.
Interesting read! Thanks for sharing.