A Veritable Guide to Who’s Who in the Artificial Intelligence Community
Having worked in a few large organizations myself, I’ve found that it can be difficult knowing who the right person to talk to is when I need something done. If you’re like me, then you probably don’t want to bother a person, especially if whatever you’re looking for doesn’t fall in their wheelhouse.
In my last post, we looked at a couple things you might ask yourself before pursuing an artificial intelligence (AI) solution. This week, we’re going to take it a step further. This guide of sorts will introduce you to the many faces that work across the AI community, including what you may expect from them as well as other names their official job roles may go by.
We have a lot of ground to cover, so let’s waste no more time with this intro and get right into it. (The roles listed below are provided in no given order.)
Ah yes, the one you’ll definitely hear most often. These folks often sit at the heart of most AI efforts because they’re the ones making the magic happen to get done what you want to get done in the form of statistical magic applied through computer code.
Data scientists will likely interact with everybody else on this list, but mostly, I’d bet they’d interact with Subject Matter Experts (SMEs) the most. This is because the best predictive algorithms have a deep understanding of what you’re trying to predict for, and data scientists don’t always have that expertise within their skillset. (We’ll cover SMEs off more down below.)
If you don’t know where to go, I’d wager that a data scientist is one of the best folks to point you in the right direction, albeit they are generally the busiest folks. If only you had a guide that would inform you other roles you might reach out to first… *wink*
Other names for this role: Statistician, actuary/actuarial analyst, advanced analytics user/consumer, machine learning analyst
Machine Learning Engineer (ML Engineer)
Of all the roles on this list, I feel that this particular one is the most confusing of all. It’s not often, but I have heard this one at times even confused with the data scientist role we just looked at. This role is well on its way to being solidified as its own, so we’ll operate with that increasingly hardened definition.
The data scientists need a place to do their work, right? When it comes to AI/ML, there are a lot of spinning plates you need to get right in order to create and execute a predictive model on the other end. This includes setting up a discovery environment, automating workflows between on premises hardware and the cloud, enabling the right things for batch transformations, setting up endpoints for continuous scoring… and the list goes on. You might be familiar with the concept of “DevOps,” so it won’t surprise you that we’ve coined a similar term for all this stuff: MLOps.
Fortunately, a lot of this is repeatable and reusable, so it makes a lot of sense for the same team to manage all these MLOps activities instead of an individual data scientist doing this over and over again. And this is where our friend the machine learning engineer comes in. Often, the ML engineer has a solid background as a data scientist themselves, so they’re intimately familiar with a data scientist’s needs and desires.
There is one tiny exception to who handles the MLOps pipeline, and that’s who deals with the data. That is handled by our next role…
Other names for this role: Technology engineer, infrastructure analyst
If you recall from any of my recent posts about data, you’ll remember that data is absolutely pivotal when creating an ideal predictive model. You can have the most glamorous AI solution in the world, but with a small amount of data or untidy/unclean data, that solution isn’t worth much. (Remember: garbage in, garbage out.)
While our ML Engineer friend is handling all the other parts of the MLOps pipeline, our Data Engineer friend is specifically handling everything to do with data. This can include making sure the data is properly stored in something like a data lake, ensuring proper data management/governance principles are applied for optimal data quality, and also working with the ML Engineer to ensure movement of data in between environments is efficient.
Data engineer tends to be a higher level role, so don’t be surprised if you see some of those “lower level” tasks farmed out to other roles. And some efforts don’t have a go-to data engineer role and instead have everything done by those same roles. I’ll leave the names of those roles below.
Other names for this role: Data analyst, business analyst focusing on data management/governance, big data developer
This role here is much more of an agile role, but I’m throwing it on this list since many companies are moving toward agile methodologies. The particular thing I want to call out is that the product owner IS NOT ALWAYS super well versed in the underpinnings that make an AI solution happen. Sometimes you’ll get lucky, and the product owner will be a data scientist or ML engineer. But what you’ll often find is that a product owner is simply a decision maker that is informed just enough to be able to make educated decisions for that particular product. If you want that finer grain of detail, you’ll probably have to go to one of the other aforementioned roles.
Other names for this role: Lead (fill in the blank)
Of all the roles on this list, I actually see this one least often amongst AI efforts, but I have a hunch this will change in the near future. When you see “UX”, you might automatically jump to thinking that this is the person suggesting the optimal user interface layout on something like a screen, and you’d only be partially correct. UX — shorthand for “User eXperience” — encompasses everything about what makes an experience the most optimal for the end user, who is often your customer.
So yes, this often comes in the form of a user interface on a smartphone screen, but AI is rapidly evolving beyond the screen. Think about self driving cars, for example. You might operate with a screen to an extent, but there’s a lot about the self driving car experience that has nothing to do at all with a screen. Like… is a car supposed to pop the trunk automatically when it rolls up so you can put in your groceries? Or what if a car tries to pick you up in a time limited area (like an airport) and you don’t show up in time? Is it supposed to automatically avoid the tow truck??
We’re entering in a really interesting time, and those aren’t questions I’d expect a typical data scientist to be able to answer. So folks, you’ve heard it here first: expect UX architect to become a more prevalent role in the AI community.
Other names for this role: Experience researcher
Ye poor software developer. How often we abuse your title! Truthfully speaking, I could have easily added “software developer” to the “other names for this role” list in almost every listed role. We’re pretty bad about slapping that role on anything we’re uncertain about.
Still, I wanted to call it out as a separate role on this list because I do believe these folks often serve a function that is not done by any other role on this list. Namely, I’m talking about the actual application your AI solution will be housed in.
So, for example, let’s say you’re building an application that algorithmically predicts the cost of your house based on a number of factors you’ve inputted as the user. The algorithm returns a value for the house by comparing those inputs to other similar attributes to other houses in your area. The algorithm (and supporting MLOps infrastructure) might be created by our other roles on the list, but what about the website or smartphone app? That’s where our friend, the software developer, comes in. Basically, anything that allows interaction with the algorithm via some channel is developed and created by a software developer.
Other names for this role: web developer, full stack developer, software engineer
Subject Matter Expert (SME)
Last but certainly not least, SMEs play a pivotal role in the AI development lifecycle. As we stated before, data scientists need to know the contextual in’s and out’s of the data they’re working with, and it’s not often the case that the data scientists will know this context themselves. SMEs partner closely with data scientists in helping provide them that context so that they can help craft an ideal AI solution, whether that be a more structured supervised learning model or a more hands off approach with an unsupervised learning model.
I end this post with this role because chances are that you, dear reader, are the subject matter expert. If you’re not one of the other roles above and find yourself wanting an AI solution for your business problem, then chances are that you’re in the best place to serve as that SME if it gets picked up by an AI project. Remember, AI is great, but it’s only as good as it is developed. Garbage in, garbage out. So if you want an ideal AI solution, an SME is critical in providing the right input to get the best output.