Types of Roles Required to Build World Class AI and RPA Teams
In our previous articles, we talked about different types of AI technologies and how and where they can be used in an enterprise. It is also important to understand different types of skill sets and organizational structures that can support businesses with these types of initiatives.
A typical lean organizational structure we see emerging is where independent enterprise technology optimization groups are established and are aligned with business but are governed by IT. We also see some organizations aligning AI and RPA initiatives within their broader shared services organizations. Although both models have their pros and cons, the choice depends on the organization and its culture. Larger organizations may even have multiple such groups in different silos.
We will now go into a bit more detail on different types of roles that exist within AI.
There are lots of non-qualified “Data Scientist” profiles on job sites. Just because someone has been writing SQL queries and working on BI platforms does not necessarily make them a Data Scientist. Even folks with experience in data warehousing technologies or big data platforms such as Hadoop does not necessarily make them a Data Scientist either. Here it is important to understand the distinction between a Database Developer and/or Data Engineer vs. a Data Scientist.
In our view, to qualify as a data scientist one needs to have the necessary academic or proven background in advanced statistics and applied mathematics, as well as computer science. We will spare you from technical Wikipedia definitions of data scientists however it’s important for you to understand the types of things a data scientist is expected to do:
- Create and execute strategies for analyzing data and extracting insights from large structured as well as unstructured data sets. Skills required here are the ability to query using traditional SQL as well as query big data sources such as Hadoop, etc. using Appache Hive, Stinger, etc.
- Create and execute strategies around ETL transformation of traditional as well as big data. Familiarity with ETL techniques and tools for data migration, cleansing and transformation is a must.
- Create and execute strategies around statistical data modeling and machine learning. Expert knowledge in breadth of machine learning algorithms and ability to find the best approach to a specific problem. Familiarity with several supervised and unsupervised learning algorithms such as Ensemble Methods (Random forests), Logistic Regression, Regularized Linear Regression, SVMs, Deep Neural Networks, Extreme Gradient Boosting, Decision Trees, KMeans, Gaussian Mixture Models, Hierarchical models, and time series models (ARIMA, GARCH, VARCH, etc.).
As you can see from above, the talent pool with such skill sets is a very limited and producing one or re-skilling an existing Data Engineer into a Data Scientist is not an easy, if not impossible, task.
Most of the talented Data Scientists usually find jobs in large companies such as Google, Facebook, Amazon, etc. so one strategy in addition to looking for lateral hires is to look for Masters or PHD graduates from some Universities which have very strong programs around Data Science. As an evolving but very important discipline, we feel that investing in the right early stage talent can pay big dividends over a period of time.
Machine Learning Engineers
A Machine Learning Engineer to a certain extent has the same level of skills as a data scientist but may not necessarily have the academic background of a data scientist. It is an important skill set which may be more readily available in the market. A Machine Learning Engineer needs to have the following skill sets:
- Solid experience with traditional SDLC (Software Development Life Cycle) and programming with traditional as well as ML friendly programming languages (Python, R, etc.)
- Familiarity with probability and statistics and understanding of some of the machine learning algorithms (these folks may be tasked to select appropriate algorithms for specific problems but may not need to understand the inner working of the algorithms in depth)
We believe that while some of the traditional Developers can re-skill themselves to become Machine Learning engineers but not everyone will be able to do so. To be a Machine Learning Engineer, the skills require some relevant academic background, aptitude and the talent and learning curve is a bit steep.
Unfortunately, technology evolves quite fast and the skills of yesterday although helpful may not necessarily translate into the experience required to move into these newly evolving engineering disciplines
A lot of traditional developers have updated their LinkedIn profiles as Machine Learning Developers. Although it’s certainly admirable that folks are up-skilling themselves, we advise that if you are looking to hire Machine Learning Engineers, you go through a thorough vetting process to qualify them.
If history is any guide, some of the COBOL Developers did not make a successful transition to GUI based application development, and many GUI based application developers could not make successful transition to mobile and social application development, not because they were not smart, but because they were stuck in maintaining legacy code bases. In addition, the developer needs to educate themselves on the math and statistics that forms the basis of machine learning algorithms.
So just because a developer can train and consume a Chatbot in their application with a simple API, does not mean that they are a Machine Learning Engineer.
RPA Architects and Engineers
RPA Architects and Engineers usually either come from a QA automation background or traditional development background. RPA is a fairly advanced area with plenty of packaged enterprise software offerings (UIPath, BluePrism, AutomationAnywhere, etc.). The RPA Engineers need to be familiar with not only the RPA software but must also be well versed with automation strategies, DevOps and infrastructure related issues that come up with any RPA program that is designed to scale.
The role of RPA Architect is an advanced role for someone who has extensive technical architecture background and has a thorough understanding of setting up Centers of Excellence for RPA Programs.
Traditional Business Analysts have always been more successful by using their business subject matter expertise along with their data analysis skills. We recommend that Business Analysts should take basic trainings on AI and machine learning technologies and what they can do for business. This will allow them to adapt quickly to the endless applications of AI and ML technologies.
Business Intelligence Developers
We see a great opportunity for BI Developers who are able to up-skill their strong data analyzing skills using ML and AI. BI Developers, Data Scientists and Machine Learning Engineers are at the core of solving some of the fundamental prediction and forecasting problems for businesses.
Project Managers, Executives & Managers
We believe that Project and Program Managers and Executive Suite folks should take strategy courses in Artificial Intelligence, Machine Learning and Robotic Process Automation. Only by taking the time to understand these technologies and their implications on business can they be ready for the hyper completive business environment which is rapidly evolving in front of us. The 4th industrial revolution is here.
We hope this article was useful for you to understand the landscape of skill sets required for AI technologies. There are many more roles that exist within the realm of AI which are beyond just the technical and business roles identified above. For e.g. the role of freelance AI trainers who are available on sites like Amazon Mechnicalturk and Crowdflower.