This year, a Stanford University course taught by undergraduate student Chip Huyen — TensorFlow for Deep Learning Research — received a surprising 350+ applications for only 20 spots. Other AI-related courses, such as Natural Language Processing with Deep Learning, and Convolutional Neural Networks for Visual Recognition, are expanding capacity to accommodate escalating student demand.
As more companies bet on AI, computer scientists with AI expertise are coming into demand on the job market. Tech giants like Google, Baidu and Amazon are fueling an AI talent race with exponentially increasing investment in recruiting engineers and researchers. At this stage, companies with more than 10,000 employees are hiring the most AI talent.
Meanwhile, more non-tech companies in sectors such as medical and health, financial services, farming, retail, and legal services are enabling AI technologies in their products and services. Gartner analysts predict software vendors will apply AI technologies into every product and service by 2020, echoing Andrew Ng’s forecast: “I actually have a hard time thinking of an industry that I don’t think AI will transform in the next several years.”
The short supply of AI experts relative to demand means that tech companies such as Google and Facebook — and even non-tech companies like Citadel and Morgan Stanley — are offering big paychecks for the right people. “The average salary, including base, bonus, sign-on, and equity for top AI talent is US$314,000 and can go into the millions of dollars per year,” says Chris Bolte, co-founder and CEO of career advising platform Paysa.
The higher end of the pay scale belongs to principal scientists and engineers with 10+ years of experience, who are generally leading project design and development. But only five percent of job listings are aimed at such talents. Says Michael Kehoe, co-founder of Silicon Valley-based education company BitTiger: “Those jobs are rare because with many applications you really don’t need too many people working on the core models. It’s too many cooks in the kitchen.”
AI-related jobs expected to grow exponentially in the near future include software engineers, who adjust models for different client applications; and data scientists, who feed the models with clean and consistent data.
Take the example of an investment company eager to embrace AI. Once they have built their core machine learning model, they will apply it to their oil investments, e-commerce investments and so on. However, different investment types have different parameters. As such, software engineers who understand specific models and can apply their expertise to software-human interfaces will be in demand.
With industry leaders like Amazon open-sourcing or licensing their machine learning models, it is possible that mid-sized companies might not need principal scientists at all, only software engineers.
Currently, there is not a specific job description for software engineers specializing in AI. As Kehoe points out: “You google ‘AI engineers’, and you’ll only see small number of senior positions. This confuses people who assume a trending topic like AI would have more open positions. The key is that many of the entry level positions are supporting engineering functions that won’t be as obvious to find using AI-related key terms.”
Another growing AI employment opportunity is for data scientists who understand the specific requirements of different industry sectors, have hands-on industrial experience with big data processing, and are well-versed in machine learning, information retrieval, and applied statistics.
Machine learning models rely on huge amounts of data. However, the data has to be pre-processed and cleaned to represent the data’s distinct attributes. For example, after a stock company draws stock information from different sources, data scientists reformat the data before it is input into models.
Rapid advancements in data processing technology will drive demand for data scientists — a recent IBM report predicts the number of data scientists will grow by more than 28 percent in the next three years.
To gain a competitive advantage in the job market, a degree in computer science from a top-ranking universities is naturally desirable. Paysa estimates that about 35 percent of currently available AI-related positions require a PhD, while over 25 percent require a master’s degree and 18 percent a bachelor’s. Some 20 percent of positions are focused on skills and experience and do not necessarily require a degree.
Having the right skill set can be as important than graduating from a specific degree program or university, particularly in the areas of machine learning, natural language processing, computer vision, deep learning, neural networks, and reinforcement learning.
Many companies are now accepting certificates earned from online courses. Last month, Andrew Ng’s startup Deeplearnig.ai introduced a series of courses on Coursera. Ng says those who earn a Deep Learning Specialization Certificate can be comfortable putting ‘deep learning’ on their resumes.
Companies will consider the breadth of an individual’s abilities as well as what AI skills are required for a specific position. Adam Curtis, CTO of SkyMind, an AI company supporting open source deep learning frameworks, says AI companies typically fall into one of two categories: R&D or product. “Most R&D companies need talent who understand math more than real world products,” says Curtis. “At SkyMind, we tend to hire engineers with strong product intuition.”
“In general we don’t look at academic papers. We look at what people have actually built,” says Curtis. “Typically, larger companies like Facebook and Google also hire more software engineers than mathematicians.”
With AI already disrupting traditional labour markets, emerging opportunities in computer science and machine learning may be some of the last jobs humans will create. It’s natural that smart humans are seeking the skill sets they’ll need to get into AI: If you can’t beat ’em, join ’em!
Journalist: Tony Peng | Editor: Michael Sarazen