“Of course demand outweighs supply. And things are not getting better any time soon. It takes many years to train a Ph.D.”
2019 AI Talent Landscape
2020 AI Talent Market Trends
Recommendations for Talent Leaders
Recommendations for AI Talent
The AI landscape evolved at breakneck speed in 2019. With AI technology predicted to add $15.7T to the global economy by 2030, 2020 is shaping up to be another pivotal year for AI research, development and commercialization.
With the tremendous growth of AI companies over the past few years, and VC investment at an all time high, there is an increasing demand for AI talent to drive innovation. An active and thriving AI talent ecosystem across all industries is critical to enable production and commercialization of AI technology at scale.
This report provides an in-depth discussion of the current AI talent landscape in the US and new trends in the talent market in 2020. We aim to support talent leaders to build strong AI teams and help AI talent thrive in a sustainable ecosystem in 2020 and beyond.
Demand for AI talent continued to rise over the past years. In the United States, AI and Machine Learning roles grew by 74% annually between 2016-2019 according to LinkedIn. Machine Learning Engineer, Deep Learning Engineer, Data Scientist, Computer Vision Engineer, and Algorithm Developer are among the most sought-after AI positions between 2018-2019, according to Indeed.
It is worth noting that the annual growth rate of AI job postings has slowed down from 136.3% in 2017 to 57.9% in 2018, and 29.1% in 2019, according to Indeed. We also observed a structural change in talent demand from massive quantities to more experienced, specialized individuals. Our AI startup talent leader survey reveals 80% of employers are looking for AI-majored graduates from top-tier schools, and 70% are looking for candidates with 3-5 years’ experience.
In November 2019, we surveyed over 80 AI startup talent leaders in Silicon Valley to understand their hiring practices and market perspectives. Most of them indicated a hiring focus on building the algorithm, research, and infrastructure teams to develop a solid technical advantage in the market. Below is an illustration of the generic AI team structure derived from discussions with AI talent leaders.
Generic team structure derived from 500 companies across industries. Data Team is responsible for organizing, managing and analyzing the data. Core AI Algorithms Team brings strong engineering capacity to set the foundation for the AI to work. As teams grow and the company matures, a Core AI Research Team may be formed especially for large tech companies. The Core Platform Team implements, maintains, secures and scales the key infrastructure. A Hardware Team is created for companies building out a physical product. AI Application and Product Team builds and manages products in a lifecycle. The Commercialization Team consists of essential business, marketing, and legal roles to companies to ensure a successful and smooth market launch and adoption.
The 2017 Tencent Holdings’ estimate of 300,000 AI practitioners and researchers worldwide provides a broad base that includes students in academia and all members of the technical teams in AI companies. The 2019 Global AI Talent Report estimated a total of 36,524 AI experts based on an analysis of LinkedIn Profiles, requiring a Ph.D. in relevant fields, solid technical skills, and at least three years’ work experience.
At the global scale, the US is leading the way in attracting, educating and retaining AI Talent, with the highest quality of AI research. However, restricted immigration policies are raising concerns for keeping highly-skilled international talent. The European Union has a comparable AI talent pool to the US, but is not fully captured by top businesses as AI investment and development of AI firms are falling behind. China has a shortage of elite AI talent as a result of brain drain, but it is catching up quickly with increasing investment in AI education and a fast learning pace.
The AI talent workforce has been growing from both online and academic training. Registrants to AI and Machine Learning degrees on Udacity reached a peak of 12,500 near the end of 2019, and the top five skills that grew in popularity from 2016 to 2019 on Udemy were all AI-related. The number of students enrolled at Stanford’s Introduction to Artificial Intelligence course grew five-fold from 2012 to 2018, and enrollment to Introduction to Machine Learning course grew twelve-fold in the same period. Applicants for the doctoral degree in Electrical Engineering and Computer Science at the University of California, Berkeley, grew from 300 in 2009 to 2,700 in 2018, with over 50% of applicants interested in pursuing AI fields. In 2018, MIT announced a new AI college on the integrated teaching of computer science with fields like biology, chemistry, politics, history, and linguistics; and the world’s first AI university opened in Abu Dhabi in 2019.
We’ve seen a rising base of AI talent from top AI conferences. The largest AI research conference, Neural Information Processing Systems (NeurIPS), hosted 13,000 attendees in 2019, a 40% increase from 2018. Over 9,200 professionals working in computer vision and pattern recognition participated in CVPR 2019 conference, up by 34% from 2018.
Although the AI talent workforce grew by 66% in 2019 over 2018, AI-related job postings are still three times that of job searches with a widening divergence.
Based on our 2019 startup Al talent leader survey, the top 3 sources to successfully recruit AI talent are from: in-house recruiters, 3rd party recruiting agencies, and job posting platforms.
This concurs with a survey of 16,000 data science and machine learning professionals by Kaggle that suggested working with recruiters is most helpful to get employed. AI and machine learning is a niche talent market that requires a solid understanding of domain knowledge and industry landscape. Working with specialized recruiters is an effective approach of talent sourcing for both employers and candidates.
Conferences and campus recruiting are secondary sources, voted by 25% - 35% of talent leaders as they are more effective for junior candidates than experienced, passive candidates.
“Of course demand outweighs supply. And things are not getting better any time soon. It takes many years to train a Ph.D.”
The rapid technology growth and the gap between AI talent demand and supply create a highly competitive landscape. Led by Amazon, Microsoft, Apple, Google, Inuit and Facebook, U.S. companies across all industries are investing $1.35 billion in AI talent, according to a 2017 survey. Experienced AI talent with specific technical skills and domain knowledge are often chased after by tech giants, posing a special challenge for startups.
Geographically, San Francisco Bay Area stands out as the highest paying market with an average base salary of $168K/year for AI-related engineering roles based on our data of 1500 voluntarily submitted job offers. Major metro areas, such as New York, Los Angeles, Boston, Seattle, offer a base salary between 150-160K. Tech hubs in Canada (e.g., Toronto and Vancouver) have a relatively lower AI talent cost compared to major U.S. cities. Considering the cost of living, Austin, Seattle, Denver and Phoenix provides the highest relevant salary, making them ideal for AI engineers to relocate to. Additionally we see AI ecosystems developing around many educational institutes such as Pittsburg with Carnegie Mellon University and Ann Arbor with University of Michigan.
Competitive AI engineering compensation package in Silicon Valley
As the #1 hub of the AI industry, San Francisco Bay Area attracts top AI talent across the globe. Built on our extensive experience working with over 500 tech companies and 15K candidates in this niche market, we provide an in-depth analysis of the compensation packages across different companies, using data from 1000+ voluntarily reported AI engineering job offers:
Growth stage startups experiencing rapid expansion and scaling (series A or B and beyond) typically offer the highest base pay, comparable to top tech companies (Google, Facebook, Apple, Amazon, LinkedIn, Uber, etc.). The median base pay for these startups is around $200K, and may even go up to $350K or higher to attract talent from tech giants for important leadership roles.
Mature tech companies established for 10+ years and companies in non-tech industries (finance, retail, healthcare, traditional transportation, etc.) offer a lower base of $150K on average. Companies in non-tech industries have limited stock options, often providing a bonus of up to 50% to attract and retain talent.
Early stage startups offer a wide range of base pay from $110-320K depending on the funding status. However, early and growth stage startups often provide significant share options to compete against tech giants. The amount of shares offered and value depends on multiple factors such as the joining time and startup stage of growth, experience level and overall company evaluation. About 50% of AI startup talent leaders in our survey indicated competitive stock options are effective for attracting and retaining talent.
Sign-on bonus, a common element of offer packages, can even ramp up to 200K when tech giants compete for top AI talent in specialized domains.
What factors influence the compensation package?
Given the competitive AI talent market compensation packages need to be carefully designed in order to appeal to candidates. Considerations to the current stage of life and career journey along with professional motivations all need to be taken into account.
Work experience in relevant AI fields significantly drives base salary. On average, we found one additional year of experience typically increases the base pay by $10K. The gradient peaks at 10 years, where most senior tech leaders cluster.
While Carnegie Mellon University, MIT, Stanford, and Berkeley are top ranked in the AI fields, AI talent in the market comes from a wide range of universities featuring strong engineering programs. Leading examples include University of Southern California, University of Michigan, Arizona State University, Georgia Institute of Technology, University of California, San Diego, Cornell University, and Purdue University.
In addition to computer science and electrical engineering, we’ve also seen high-quality candidates transition from majors like mechanical engineering, physics, mathematics, statistics, bioengineering, and aerospace engineering.
An AI engineer with a PhD degree in related fields can earn $20K-40K/year more than those with a master’s degree. While most AI engineering talent have a graduate degree, intelligent and motivated candidates with a bachelor degree can also be competitive; solid hands-on/internship experience makes them on par with master’s degree postgraduates.
High-level technical leaders (e.g., senior directors, VPs, etc.) are the highest paid AI roles, and would earn $200K/year more than junior ICs (individual contributors). Mid-level leadership roles (engineering managers, directors, etc.) and senior ICs earn $50K and $20K more on average than junior ICs. Stock shares typically grow with seniority and high-level management roles may have a higher stock value than their base salary.
Our analysis shows the salary for women and men is not significantly different for AI engineering roles if their education background, experience and seniority is equal. The scarcity of AI talent is likely to attenuate the impact of gender bias.
In the competitive market, it typically takes 1-3 months to hire depending on the company and the hiring process. Junior positions typically interview 3-10 candidates and senior roles mostly interview 10-20 candidates. Candidates we work with who are actively looking typically have 3-5 offers; passive candidates typically have 1-2 offers.
“This is a different generation of talent in today's AI market. They have more choices. They care more about perks and the challenge. What's the career progression? How does my role influence decision making?”
The shift in the industry landscape makes it much more challenging and strategic to acquire talent. The conventional approach of information based-recruiting with only information about the employer and position is not going to work with the new generation of the workforce. Alex Ren, founder of TalentSeer & Robin.ly and AI investor, shared an innovative “3i Recruiting Mindset” that has been proven effective in today’s AI talent market. This approach involves a more in-depth engagement with candidates to discuss the industry, business, and career insights; amplifying employer & personal branding through social networks.
TalentSeer 3I Recruiting Mindset. Insight-based and influence-based recruiting are elevated from commonly used information-based recruiting. Sharing in-depth insights on industry, business, and career development through various platforms will help amplify the impact at scale and build thought leadership.
While large tech companies often offer very competitive salaries, startups in Silicon Valley often rely on other approaches to attract talent based on our AI Talent Leader Survey:
“The number one reason people leave companies: people leave people. People leave their manager way more often than they actually leave the company.”
Retaining talent is equally important for AI companies and teams given the competitive market and highly motivated talent pool. Below are top strategies to retain talent summarized from our conversations with talent leaders and candidates.
AI talent is solving some of the most cutting-edge technology problems whilst constantly breaking new grounds. Keeping your team intellectually challenged is important to retain the talent. However, technical/data problems that are ever-changing and cannot be solved in a foreseeable timeframe are one of the top reasons that AI talent chooses to leave.
AI talent often leaves because of the lack of impact, either within the company or in a broader societal landscape. “Results not used by decision-makers” was voted as a major barrier at work by 25% data and machine learning professionals in a Kaggle survey.
Clear communication is the key to a functional team. Elaborating on the context and background of decisions, along with the decision itself, is helpful to align the team and eliminate unnecessary tensions.
The top two technical challenges for AI talent are the lack of strong engineering teams to build infrastructure, and the lack of data to train machine learning/deep learning models. Without the foundational support on data and infrastructure, AI talent is bound to look for other opportunities to maximize their productivity.
While the fields of AI and Machine Learning are advancing, there is still a significant lack of diversity within the industry with regards to education, gender, ethnicity, experience, and other relevant parameters. Only 18% of researchers publishing at the top AI conferences are women. Only 15% of AI research staff at Facebook are women, and only 10% at Google. African American workers appear to only represent 4% at Facebook and Microsoft respectively, and only 2.5% at Google.
This likely has a negative effect on the types of programs and systems being created, causing them to be built with biases and discrepancies towards minorities. The National Institute of Standards and Technology recently tested 189 major commercial facial-recognition algorithms and found they falsely identified African-American and Asian faces 10-to-100 times more than Caucasian faces. It is also reported that Apple’s credit decision algorithm offers lower credit limits to women than to men. A diverse background of people should be involved in AI research and development, especially when they are going to be used by (and on) the general public on a global scale.
2019 saw increasing commitments to improve diversity and inclusion in the AI field. More and more AI companies are seriously incorporating diversity in their hiring process and even setting quotas for underrepresented gender and ethnic groups. The AI companies we partner with are no exception to this. The world’s largest AI conference NeurIPS not only hosted more diversity-related workshops in 2019 (e.g., Women in Machine Learning, Black in AI, and LatinX in AI), but also changed its acronym and logo to be more inclusive and unbiased.
From password-free voice authentication, to humanoid service robots and intelligent health tracking, numerous consumer-facing AI products emerged at CES 2020. At the same time, AI solutions are becoming mainstream in boosting business efficiency and productivity. The next wave of AI production and commercialization is approaching in 2020.
Enterprises are keen to implement AI solutions. Nearly 80% of chief information officers (CIO) in U.S. companies plan to increase the use of AI and machine learning in 2020. However, only 14.6% of industry leading firms have actually deployed AI capabilities into widespread production. The gap between expectation and implementation yields great potential for more progressive AI production and commercialization in 2020.
Our survey of AI talent leaders in Silicon Valley startups supports this statement. 40% of respondents listed “developing products from R&D” as their primary business direction in 2020. Consequently, these companies are expecting to scale their product teams (front-end engineers, user experience experts, product managers, etc.) and commercialization teams (business development, marketing, and legal professionals) in 2020 for successful market adoption. Experienced product leaders who understand both the technology and business aspects are especially on demand.
“It's not just a technology problem, it’s also a ‘people in a process’ problem.”
Talent Leader Recommendation #1
Companies developing AI products should focus on building a stronger team structure and searching for product leaders. Successful product development requires holistic team dynamics and workflows. It is important to ensure everyone, from the data scientist and machine learning engineer, to marketing and sales, is aligned with the business goal.
AI Talent Recommendation #1
Engineers interested in making a direct impact on production should start honing product management and communication skills, and developing a better understanding of the targeting market through online training and practices at work. It is also helpful to learn from product leaders and engage with the targeting end-user community through online and offline events.
As AI technologies mature, we are expecting applications proliferate across a broad range of industries, from intelligent chatbots for banking and financing, to translatable solutions across health systems, and streamlined digital farming systems combating climate change.
AI talent that used to focus on autonomous driving and robotics are more open to opportunities in non-tech industries (finance, healthcare, retail, agriculture, etc.) to bring innovation closer to our everyday life. Candidates with relevant industry background possess an advantage in the market compared to peers with similar AI and machine learning expertise.
Talent Leader Recommendation #2
Non-tech companies seeking experienced AI engineering talent to lead AI transformation should be prepared to match the competitive market compensation. Alternatively, talent who has transitioned to AI with experience in related industry domains can potentially be an affordable workforce.
AI Talent Recommendation #2
Moving into a new industry can make a significant impact on long-term career paths and professional networks. It is critical for AI talent to conduct thorough research about the new industry and carefully evaluate the transition.
Candidates working in specific industry domains with training from indirect AI fields (e.g., physics, statistics, semiconductor) would also have a good chance to become an AI engineer if well prepared. A good stepping stone is to enter an AI company with their current expertise and then strengthen their AI and machine learning skills in an immersive environment.
“Talent exists everywhere, and increasingly so does opportunity, so smaller tech hubs are giving Silicon Valley the fiercest competition of its existence.”
The shortage of AI talent and escalating salary requirements is making the labor cost and global reach a significant burden for many companies, especially in Silicon Valley, where most AI companies are clustered. Tech giants such as Google, Microsoft, Amazon, and Facebook have already tackled the problem by opening their own AI labs in places outside of the US, such as Canada, India, South/East Europe, and China, where competition for AI talent is less fierce. Building remote teams also provides a global reach for these companies.
30% of Silicon Valley-based AI startups in our AI talent leader survey also indicate a plan to expand the team remotely in 2020. In addition to the overseas technology hubs, low-cost areas in the U.S. with access to university graduates, such as Michigan and Texas, are also favored by companies for easier team management. Canada is another appealing location, especially for attracting international talent with its amicable immigration policy.
Talent Leaders Recommendation #3
Companies can consider exploring the AI talent market and remote workforce in alternative tech hubs. To prepare for the distributed teamwork mode, it is important to start adopting technology and processes for team integration, such as video calls, group chats, and work-flow synergies across time zones.
AI Talent Recommendation #3
Considering Silicon Valley has a high cost of living, AI talent may look to other locations in the United States, Canada or alternative tech hubs for better quality of life.
While technical skills within AI are fundamental, tech giants and startups are also highly interested in individuals who have non-technical skills to match. The 2019 LinkedIn Global Recruiting Trends Report reveals that 92% of talent managers found strong soft skills are increasingly important for business success.
In particular, creativity, critical thinking, growth mindset, resilience, and communication are among the top non-technical skills on demand by AI companies and prominent in the 2020 workplace learning trends. AI talent today is facing unprecedented technical challenges and substantial ambiguity, thus it is crucial to have an open mind, innovate at a fast pace, and stay resilient to uncertain conditions. Effective communication is also important to ensure smooth collaboration.
Recommendations for Talent Leaders #4 Building the right culture of innovation and responsibility is key, especially for startups, where everyone is expected to be hands-on in the early stage. Encouraging individuals to take ownership of their work will allow each member to grow much faster into an integral part of the team.
Recommendations for AI Talent #4 Professionals currently working in or preparing to enter the AI space should evaluate their non-technical strengths and potential fit with the company culture. Expressing these non-technical qualifications during interviews would give them an edge in the competition.
We’ve seen a growing number of students enrolling in bootcamps and online AI courses. As the most popular AI e-learning course, “Machine Learning” offered by Stanford University had over 2.7 million enrollments by the end of 2019. The entire e-learning market is expected to rise over $300 billion by 2025. The US/Canada-based in-person coding bootcamps have 17.5 K graduates in 2019, with an accelerating growth rate year over year.
Statistics show these types of training to be useful for career advancement -- 76% software engineers found bootcamps helped prepare them for relevant jobs. Additionally, 57% of employers said they would hire bootcamp graduates, with 36% saying they were unsure. It should be noted that lack of experience is the main reason candidates get rejected, particularly for startups, according to responses from our AI talent leader survey.
Talent Leaders Recommendation #5 Explore a broader base, including bootcamps and community colleges, for talent resources. They can also utilize online training to upskill current technical teams. A tiered team structure with both experienced leads and young professionals is a strategic approach to balance cost and productivity.
AI Talent Recommendation #5 Experience is key. Bootcamp/nano-degree students need to be exposed to real world projects and gain experience to augment online training and mentorship. They must think creatively about opportunities to get involved, especially with early-stage companies to gain experience, such as via pro-bono work.
As AI matures in 2020 there will continue to be a great demand and limited supply for talent and this will remain for 3-4 more years and the education funnel catches up. As AI matures, and as teams evolve, research and commercialization will speed up the need for the additional team members and critical functions such as Sales, Marketing, Business Development, Product Management etc. For these roles additional skills and understanding of data and AI will be required in order to best present these innovative technologies to the market.
This report is based on TalentSeer proprietary information including 1500 voluntarily reported job offers from the US and Canada (1000 job offers in San Francisco Bay Area), 80 responses by executives and talent leaders from AI startups in Silicon Valley, and daily engagement with over 500 tech companies and 15K candidates. Our data analysis focuses on AI engineering roles in the San Francisco Bay Area.
TalentSeer is the fastest-growing AI talent partner in the U.S. providing integrated talent acquisition, market research, and employer branding services. With an engaged AI community - Robin.ly, an innovative AI recruiting approach, and deep domain knowledge, TalentSeer has helped with 150 partners across autonomous driving, internet, finance, retail, and healthcare industries to build strong AI teams.
Learn more about AI talent market and talent recruitment support, contact us at email@example.com