An effective enterprise AI team is a diverse group that includes more than just data scientists and engineers. A successful AI team should also include a lot of people who understand the business and who are trying to solve problems, said Bradley Shimmin, principal analyst for AI platforms, analytics and data management at consulting firm Omdia.
“The technology and tools at our disposal increasingly require us to support and empower those domain professionals, business users or analytics professionals to use and take responsibility for AI directly within the company.”
Carlos Anchia, co-founder and CEO of AI startup Plainsight, agrees, arguing that success in AI depends largely on building a well-rounded team with a variety of advanced skills, but it is very challenging.
“It may seem easy to be clear about what makes an effective AI team, but when you look at the detailed responsibilities of everyone on a successful AI team, you quickly conclude that building such a team is very difficult ,” he said.
To help you build an ideal AI team, let’s take a look at these 10 key roles that should be on that team.
data scientist
Data scientists are arguably the heart of any AI team. They process and analyze data, build machine learning (ML) models, and draw conclusions that can be used to improve ML models that are already in production.
Mark Eltsefon, a data scientist at TikTok, said a data scientist is a mix of a product analyst and a business analyst with some machine learning knowledge.
“Their main goal is to understand which key metrics have a significant impact on the business, collect data to analyze possible bottlenecks, visualize different user groups and metrics, and propose and develop solutions on how to improve those metrics ,” he added that when developing new features for TikTok users, it’s impossible to understand whether the feature benefits or alienates users without data science.
“You don’t know how long you should be testing features and what exactly you should be testing, and for all of those things, you have to use an artificial intelligence approach.”
machine learning engineer
Data scientists can build machine learning models, but it is machine learning engineers who implement those models.
“The job of a machine learning engineer is to package a machine learning model into a container and deploy it into production — usually in the form of a microservice,” said Dattaraj Rao, innovation and R&D architect at technology services firm Persistent Systems.
The role requires specialized backend programming and server configuration skills, as well as knowledge in containers, continuous integration and delivery deployments, Rao said. “Machine learning engineers are also involved in model validation, A/B testing, and production monitoring.”
In a mature machine-learning environment, machine-learning engineers also need to experiment with serving tools, which require only a small amount of experimentation to find the models that perform best in production, he said.
data engineer
Data engineers build and maintain systems that make up an organization’s data infrastructure. Erik Gfesser, director and chief architect at Deloitte, said data engineers are critical to AI initiatives because data needs to be collected and made suitable for use before it can be used to do other valuable things.
“Data engineers build data pipelines to collect and aggregate data for downstream use, and in a DevOps environment, they build pipelines to implement the infrastructure that runs those data pipelines,” he said.
Data engineers are the foundation of both machine learning and non-machine learning projects, he said. “For example, when implementing a data pipeline in a public cloud, data engineers need to first write scripts to start the necessary cloud services, which then provide the computations needed to process the ingested data.”
If you’re building a team for the first time, you need to know that data science is an iterative process that requires a lot of data, says Matt Mead, CTO of IT services firm SPR. Assuming you have enough data, “About 80% of the jobs will be data engineering related, and about 20% will be practical jobs related to data science.”
Because of this, very few people on the AI team work in data science, he said. “The rest of the team needs to identify the problem being solved, help interpret the data, help organize the data, integrate the output into another production system, or present the data in a presentation-ready way.”
data administrator
Data stewards oversee the management of a company’s data, ensuring that it is accessible and of high quality, an important role that ensures that data is used consistently across the organization, and that the company complies with changing data laws.
Ken Seier, head of data and AI country practice at technology firm Insight, said data stewards want to make sure data scientists have accurate data and that everything is repeatable and clearly marked in the data catalog.
People in this role need to understand data science and have the communication skills to collaborate across teams and work with data scientists and engineers to ensure data is accessible to stakeholders and business users.
Data stewards also enforce the organization’s policies regarding data use and security. “Data stewards want to make sure that only people who should be given access to secure data have access to it,” Seier said.
domain expert
Domain experts have an in-depth understanding of a specific industry or subject area, are an authority in a certain field, can judge the quality of available data, and can communicate with the intended business users of an AI project to ensure that the project has real-world value.
Domain experts are essential, because few technologists developing AI systems have expertise in the system’s target domain, said Max Babych, CEO of software development firm SpdLoad. “Domain experts can provide critical insights that allow AI systems to perform at their best.”
When SpdLoad developed a computer vision system to recognize moving objects on autopilot as an alternative to LIDAR technology, they started the project without domain experts. What SpdLoad didn’t know, despite studies proving the system worked, is that car brands prefer lidar over computer vision because of the technology’s proven reliability, and they don’t have the opportunity to buy a computer vision-based product .
“One of the key pieces of advice I want to share is that you think about the business model and then bring in domain experts to see if that’s a viable way to make money in the industry, and then move on to the more technical issues.”
Domain experts can be important liaisons between clients and AI teams, said Ashish Tulsankar, head of AI at education technology platform iSchoolConnect.
“This person can communicate with customers, understand their needs, and provide the AI team with a series of next steps. And domain experts can also oversee whether the company is implementing AI in an ethical manner.”
AI designer
An AI designer is responsible for working with developers to ensure they understand the real needs of human users. This role envisions how users will interact with AI, creating prototypes to demonstrate use cases for new AI capabilities.
AI designers also ensure that trust is established between human users and the AI system, ensuring that the AI can learn and improve from user feedback.
Shervin Khodabandeh, co-leader of BCG Boston Consulting Group’s North American AI practice, said: “One of the difficulties organizations have in scaling AI is that users don’t understand the solution, don’t identify with it, or can’t interact with it. Those who are getting value from AI organization, and their secret sauce is actually that they can do human-computer interaction in the right way.”
Boston Consulting Group follows the 10-20-70 principle: 10% of the value is algorithms, 20% is technology and data platforms, and 70% is business integration, or tying it into corporate strategy in business processes.
“Human-machine interaction is absolutely critical and a big part of 70 percent of the challenges,” he said, adding that AI designers will help you achieve your goals.
product manager
Product managers are responsible for discovering customer needs, responsible for product development and product marketing, while ensuring that the AI team makes favorable strategic decisions.
“In an AI team, the product manager’s role is to understand how AI can be used to solve customer problems and then translate that into product strategy,” says Dorota Owczarek, product manager at AI development firm Nexocode.
Owczarek was recently involved in a project to develop an AI product for the pharmaceutical industry that would support human review of research papers and documents using natural language.
“This project required close collaboration with data scientists, machine learning engineers and data engineers to develop the models and algorithms needed to power the product,” she said.
As Product Manager, Owczarek is responsible for implementing product roadmaps, estimating and controlling budgets, and handling collaboration between product technical, user experience and business aspects.
“In this particular case, since the project is initiated by business stakeholders, it is especially important to have a product manager who can ensure that their needs are met while focusing on the overall goals of the project,” she said, adding that AI Product managers should have both technical skills and business acumen.
“Product managers should be able to work closely with different teams and stakeholders. In most cases, the success of an AI project will depend on collaboration between business, data science, machine learning engineering, and design teams.”
AI product managers also need to understand the ethical considerations associated with AI, Owczarek said. “They are responsible for developing internal processes and guidelines to ensure a company’s products comply with industry best practices.”
AI Strategist
AI strategists need to understand how the enterprise level works and coordinate with executive teams and external stakeholders to ensure the company has the right infrastructure and talent for the success of AI initiatives.
To be successful, AI strategists must have a deep understanding of their business domain and the fundamentals of machine learning; they must also know how to use AI to solve business problems, says Dan Diasio, global AI lead at EY Consulting.
“A few years ago, technology was the hard part, but now, technology is reimagining the way we connect different businesses to take advantage of the AI capabilities or AI assets we build.” He added that AI strategists can help companies to Transformative thinking to think about how to use AI.
“To change the way [companies make] decisions, it takes people with significant influence and vision to drive the process.”
AI strategists can also help organizations obtain the data they need to effectively drive AI.
“Today, the data businesses have in their systems or in their data warehouses really represent only a tiny fraction of the data they need to build their AI capabilities. Part of the role of an AI strategist is to look to the future and see how Access and leverage more data without violating privacy rules.”
Chief AI Officer
The Chief AI Officer is the primary decision maker for all AI initiatives and is responsible for communicating the potential business value of AI to stakeholders and customers.
“Decision makers are those who understand the business, the opportunities and the risks,” said iSchoolConnect’s Tulsankar.
Chief AI officers, he said, should know what human AI can be used for and which can deliver the most important economic benefits, and they should be able to articulate those opportunities to stakeholders.
“They should also discuss how these opportunities can be realized iteratively. If there are multiple customers or multiple products that require AI, the chief AI officer should be able to separate out the customer-agnostic and customer-specific parts of the implementation.”
executive sponsor
The executive sponsor needs to be a C-level executive who takes an active role in ensuring the AI project achieves results and is responsible for securing funding for the company’s AI initiatives.
EY Consulting’s Diasio said executives play an important role in helping drive the success of AI projects. “For companies, the biggest opportunities come from areas where they tend to break out of specific functions.”
For example, a consumer products manufacturer has a team responsible for R&D, a team responsible for supply chain, a sales team, and a marketing team, “The biggest and best opportunity to apply AI to transform the business is related to all four functions. , so making these changes requires strong leadership from the CEO or C-suite.”
Unfortunately, many executives at many companies don’t fully understand the potential of AI, says BCG’s Shervin Khodabandeh.
“They have a very limited understanding of AI and often see AI as a black box that’s thrown directly at data scientists, but they don’t really understand what new approaches are needed to use AI.”
Adopting AI will be a huge change in corporate culture if companies don’t understand how AI teams work, how roles work, and how they are empowered, he said. “99% of traditional companies adopting AI would think it’s a hard thing to do.”
GIPHY App Key not set. Please check settings