The role of a Data Scientist is the most sought-after position in the world of tech. In today’s data-driven business environment, organizations are looking for data scientists who can turn volumes of data into actionable insights. The niche data scientist persona is often undefined and misunderstood which creates several problems throughout the hiring lifecycle. The average time taken to hire a data scientist in the US is roughly 48 days as opposed to the 37 days it takes to fill a software engineer’s position. Higher attrition rates (14% > 9%) and lower average tenures (1.5 years < 2.4 years) are also plaguing companies looking to hire and retain Data Scientists.
Most of these problems stem from the lack of clarity in how to identify the right Data Scientist and what companies really want out of them. Identifying and alluring the right person for the job is almost always more challenging than hiring itself because data science is a highly diverse and complex pool of skillset and talent.
While the Data Scientist usually gets assigned a multitude of names across various organizations, we’ve identified four broad types which are – Data Scientist, Applied Data Scientist – NLP, Applied Data Scientist – Vision, and Data Scientist Speech Recognition. The average Data Scientist must be aware of tools such as ML Frameworks (SciKit learn), DL Frameworks (Theano), and TensorFlow, Keras, Caffe. An Applied Data Scientist and Data Scientist typically possess skills like Gurobi, CPLEX, Symphony, HTK, TTS, and OpenFST. Here is a breakdown of behavioural skills, tool proficiencies and experience that each type of data scientist is expected to possess.
A Data Scientist is ideally proficient in programming languages like R, Python and OpenCL. Some of the other IT tools and platforms a Data Scientist must know are:
• Applied AI as a Service – Image Recognition AAS, NLP AAS, Vision AAS
• Integrated AI Platforms – AZURE ML, AWS Deep Learning AMI, Google ML, DataBricks ML
• AI Frameworks – SciKit Learn, TensorFlow, CAFFE, TORCH
• Big Data Platforms – EMR, DataBricks, Spark
• Infrastructure & Processing – GPUs, CPUs, AWS Lambda
Role of HR: Creating the Ideal JD
Creating an optimal JD attracts the best talent, and HR teams must cover all bases in order to get the right fit that goes beyond an individual’s skillsets.
• Identify core skills required for the job – Statistics, ML, NLP or Computer Vision, specify the programming languages that are typically used, and define the understanding of frameworks based on the team’s tech stack.
• List the responsibilities to be taken up by the employee across key business functions – Sales & Marketing, Finance, Operations or Customer Experience, and specify certain use cases the hire is expected to work on. Accentuating on key organizational values that a data scientist would prefer working in is another point to tick.
• A culture that promotes learning and innovation, supports risk-taking behaviour, and a fast-paced environment will generally attract quality talent pool. Having a gender-inclusive environment, and using gender-neutral titles in the JD also helps.
• Use descriptive titles like “engineer”, “project manager”, and “developer” as opposed to titles that suggest masculinity. Also, avoid using gender-charged words. For example, words such as “analyse” and “determine” are typically associated with male traits, while “collaborate” and “support” are considered more feminine.