So, you are interested in AI. But what language do you learn to code? Do you learn to code at all? One of the key differences between AI Specialists and Data Scientists is there is less of a reliance on being able to write code. Before we get into the differences, let’s discuss a few of the finer points of ‘being able to code’. 

Understanding the roles

Most data sciences can write code in Python or R primarily (with some SQL thrown in for good measure), usually inside of a Jupyter notebook or similar environment. While this code is useful for developing and creating models, very rarely will it see production. Production level code needs to adhere to stricter coding standards and testing standards which most Data Scientists do not spend time on. These standards are where the differentiation between roles starts to come into play, with the expectation of coding abilities:

Machine Learning Engineers: MLE’s write production-level code that uses software engineering concepts and testing procedures for repeatable and reliable code. MLE’s have a greater focus on the production aspects of data science and work closely with software engineers and infrastructure engineers to deploy quality code for use cases. 

Data Scientists: Rarely expected to write production-level code, usually write code within isolate environments (Jupyter Notebook/Other cloud workbenches) with less of a focus on the code, more of a focus on the science behind it.

AI Specialists: AIS’s have a blended approach if you are using cloud data science platforms (BigML, H20.ai, Dataiku, etc.). You may not need to know a great deal of programming, aside from the ability to use SQL and basic Python as most of the interfaces and interactions happen with a user interface. This level of coding means the primary skills are understanding how to build and tailor models and how to apply them to business use cases. 

One of the first pieces of advice given to people aspiring to enter into the world of AI is to learn Python, which here at AI Specialist blog, we also advocate as a fantastic first step. Python enables you to get into the depths of many different models and provides high-level APIs for use with the most cutting edge modelling approaches. If a deep dive into statistics is more your thing, then R may be a better fit for your skillset. The majority of workbenches and data science platforms support both equally. 

One of the first pieces of advice given to people aspiring to enter into the world of AI is to learn Python, which here at AI Specialist blog, we also advocate as a fantastic first step.

The first question you must ask yourself is which role you want to align closer too: Machine Learning Engineer, Data Scientist, or AI Specialist, as it will change the type of code, you learn to build and understand. If you are leaning more toward the AI Specialist role, you only need to understand enough Python to get you through analyzing a Jupyter notebook. And being able to display outcomes and make sure your code is sharable (i.e., it can run when given to someone else!). The clear difference here is not learning the in’s and outs of Python to deal with things like asynchronous processing and multithreading, which are used when deploying Python as production code. You also won’t often need to ‘open up’ Python packages and modify the function of models like you would need to as a Data Scientist, which would require a more in-depth understanding of how classes and functions operate. 

What to learn

The best way to approach this is to follow some tuitions on using Python in Jupyter Notebook, how to use libraries as scikit learn, and how to interact with API’s on a fundamental level so that you can use cloud services and automated platforms with code. Data Scientists will also need to learn things such as how to use Linux operating systems and in some cases Docker containers, AI Specialists, on the other hand, shouldn’t need to be familiar with these types of services. 

The following table will help to show you the type of tools each kind of role uses; the definitions used are:

  • Expert: Uses the tool consistently, an expert in being productive with it
  • Intermediate: Good knowledge of the tool, can use it for day to day tasks
  • Beginner: Sporadic use of the basic functionality 

Keep in mind this table is limited to the actual toolsets. Not the type of modelling that is going on behind the scenes. Also, consider the vital role of the Data Engineer who has an incredibly large role to play in the feature engineering steps. 

Data Scientists will also need to learn things such as how to use Linux operating systems and in some cases Docker containers, AI Specialists, on the other hand, shouldn’t need to be familiar with these types of services. 

Using this table and the information of roles above, you can tailor your learning plan for what you need to know! Check out our Training section and Reading List section for some depth links and resources to kickstart your training. 

Leave a Reply

avatar
  Subscribe  
Notify of