Each day we come across an exponentially increasing number of AI solutions around us. While some of these solutions can be seen (like client support bots or biometric facial recognition), others stay in the back-end systems where only the most significant level of specialists can “see” and value their input. In any case, they exist, thus one cannot resist the urge to ponder – who do you really need to make an AI project successful? You can’t simply mail-request them or download from a disk. You need a particular skillset from actual humans to make it work.

The 4 Critical Roles

Subject to the complicity of the procedure you are automating, the number of actual individuals behind those roles may fluctuate, therefore, we should not think about those as FTE-s but specific jobs that should be dealt with. The minimum number of a living and breathing soul behind each role is 1 but dependent on the job to be done — it could be more.

 1. The Business Owner

This is somebody who feels that he/she is in a difficult situation. I must underline this as much as possible — it must be a real person with thoughts and feelings and emotions since organizations don’t have issues — the individuals who fill specific tasks for those organizations do. An organization is only a number in some registry, everything that creates the worth around the name of the organization is created by the individuals who make this organization work.

The good thing about an AI project is that it can’t and shouldn’t begin from IT, but from the individual (from the business side) who has an issue. What’s more, it truly makes no difference whether this individual is from the accountancy department, law or mechanics office.

This individual will presently hold first of the basic 4 jobs and it will be up to him/her to:

• Define the issue we are solving and clarify why it is essential (to whomever it might concern)
• Make the GAP analysis and draw out the (objective) contrasts between the situation today and how it should ideally look like later on
• Be prepared to put a lot of time, exertion and energy into anything that is next to come, hence understanding this isn’t simply one more IT development project where you describe your demand and hang tight for the brighter future to come while the IT wizards do their thing.
• Pick the right measurements and KPIs that ought to be estimated so as to evaluate how fruitful your AI project really is, and furthermore set targets for the entire group based on that
• Will continually (and all through the task) be the one to assess how the group is getting along (business tight clamp), giving the criticism and directing them toward the business results wanted.
• Will set up the vital procedures to keep the AI project in accordance with business after the development stage has finished.

 2. Data Engineer

No good thing can ever originate from a bad dataset. Artificial intelligence won’t be the miracle tool that transforms water into wine — for you to get the wine, you need to give your team appropriate grapes and somebody needs to develop, collect and set them up. That is fundamentally what a data engineer does.

The Data engineer will be liable for: 

• Building datasets from data sources
• Developing and dealing with the infrastructure to move the data
• Building the pipelines from (at times a few distinct sources) and setting up the data for a data scientist, which for the most part implies a multi-step procedure to construct, test and maintain architectures, for example, databases and enormous scale processing systems
• Will be “always” responsible for the quality and ease of use of the datasets utilized by your AI

It has no effect on whether every one of those assignments is taken care of by one or a few different individuals. It is perfectly typical to have (a group of) individuals who have in-depth knowledge of your data sources and the connections between IT systems, as well as the ability to ensure that those systems will continue giving the vital data.

And afterward, another group of individuals who realize how to make this all helpful for a data scientist — to assemble the pipelines and handle the technical systems for those purposes.

 3. The Data Scientist

Much the same as with winemaking, it takes far beyond simply great grapes to deliver an astounding wine, and along these lines, this is where the actual magic and mystery occur.

This is what it resembles and how it ought to be regarded by others, mere humans, I mean.  You ought to never permit any amateur close to your data. Especially those who may “have recently googled data science, read 5 blog posts and now are super-excited to give it a shot”.

The Data scientist will:

• Purify the data so it could be utilized for building a model
• Find the trends and patterns that are covered up in the data and are really significant for taking care of this specific business issue (not simply odd coincidences)
• Build and train the machine learning models that will be the heart and brains of your AI
• Evaluate the outcomes and connection with the business KPIs
• Visualize and communicate the outcomes for the parties involved

This will be a blend of mathematics, statistics, programming and even a little dash of biochemistry — so in short — not for the layman. Much the same as owning a calculator doesn’t make you an accountant, you can’t simply wake up one day and conclude that you are a data scientist.

4. The Machine Learning Product Manager

We could also consider him or her a translator, yet that has a temporary whiff to it and this position is unquestionably anything other than temporary.

This job will be the one to:

 Understand and clarify how the business units work and how business value is made

• Knows what is “conventional IT” and knows about the normal programming languages yet additionally comprehends what is AI and how it separates from regular IT
• Has great relationship building abilities and realizes how to convey to the two sides (and clarify/interpret between them)
• Owns the tool stash of driving individuals without legal authority
• Keeps the group productive, within lines and unmistakably discloses the expectations to every member

When every one of those boxes is ticked — you are good to go and all set!

The key ingredients are as of nowhere today for you to construct your first (or next) AI solution. All you need is a decent group to bring the outcomes home. Without a doubt, science will continue advancing, but there is no good reason to stop and pause while that occurs.

On the off chance that you wish to consider your company to be as prosperous as it is today in 5 years’ time, or t 5 times more successful by then — then today is the day to begin.

Characterize the business problem, discover the players for every one of the 4 roles referenced above and start testing! Simply remember that the last role is the way to everything. Else you’ll wind up with a fork, a destroyed wine bottle and cork crumbles everywhere.