OnQ Blog

How Qualcomm develops its AI and machine learning workforce

At Qualcomm, our goal is to “invent breakthrough technologies to transform how the world connects, computes and communicates.” In order to do that, we must first foster the people who work here. We do this, in part, through the education and empowerment of our employees.

Take our machine learning (ML) efforts, for example. We’re working aggressively to build teams with deep learning (DL) expertise for our computer vision, autonomy, and GPU groups as well as acquiring companies, like artificial intelligence (AI) research company Scyfer, to bolster our work in on-device AI. We’ve also found that we’re uniquely positioned to address a substantial portion of our ML talent needs by tapping into our own employee population. To that end, the company’s engineering and HR teams have collaborated to design and deploy a strategic effort to develop ML/DL skills among our existing workforce.

 
 

Here’s a look at what we’re doing to cultivate AI expertise in our own backyard:

  • The first step in this strategic learning initiative was for HR to partner with engineering leads to deconstruct ML in terms of relevant core skills and capabilities (e.g. experience with Python, linear algebra, and ML frameworks like TensorFlow, Caffe, etc). We then aggregated and mapped relevant skill clusters to discrete engineering roles and responsibilities (e.g. ML SW Apps and ML Scientist). Finally, we used an internal skills assessment process to calibrate these requirements with existing capability and discern specific areas where there were skill and competency gaps.
  • This analysis formed the basis of our custom technical training and developmental program. Over 1,000 of our engineers have gone through this developmental program in the last nine months, allowing our engineers to start learning and deploying ML based on their team’s specific needs. These needs range from getting up and running with TensorFlow/Caffe to training simple models to developing a deeper understanding of the math required for creating new algorithms for areas like autonomous driving and computer vision.
  • Our next step was to evaluate various external ML training vendors. We’re currently working with a select few to develop and deploy these programs to our global R&D centers in China, Europe, India, and the U.S., and we’re continuing to look for additional interested parties. Our employees also have access to curated ML/DL content from online educators, like Udacity, Coursera, and LinkedIn Learning (formerly Lynda.com), to supplement their learning. Tuition assistance is available, so employees can choose any relevant ML/DL program from accredited institutes globally and get fully reimbursed.
  • In partnership with our engineering leadership, we’re piloting an internal talent marketplace where various ML/DL tasks are posted. These are similar to graduate-level projects and are intended to allow engineers to transfer their learning from the classroom to the workplace. Opting in to these tasks provides engineers with access to an experienced mentor, while providing hands-on experience that allows them to become familiar with internal tools and processes.
 
 

A key aspect of this initiative is to use this analysis to help focus our recruitment strategy. We partnered with various research and advisory companies to do a workforce and location analysis of specific ML talent. This allowed our leads to make informed decisions on where to move requisitions and roles so they could get up to speed quickly, while minimizing project delays and disruptions.

Working with academia is another key cornerstone of our talent strategy. We identified professors and universities who are working on ML research areas of interest to us. Our engagement with these universities ranges from charitable giving to guest lectures to faculty sponsorship awards and innovation challenges. In addition, we developed a new “Research Associate” position within engineering: This is a targeted 12-month role intended for recent ML/DL PhDs and Postdocs. The objective is to provide these new grads with a low risk, seamless transition into the industry, and it often allows them to extend research they were working on in school. These associates are currently located in San Diego and India, and we’re looking to expand the scope this coming year.

The market for ML talent is heating up, and this has some companies embarking on a “free agency” approach to poaching talent. That’s why doing this analysis to develop talent “at home” is so critical. We’ve found that a strong organizational culture that encourages talent mobility and internal redeployment and investment in customized learning and development initiatives is a worthwhile approach to creating and sustaining an engaged workforce.

Opinions expressed in the content posted here are the personal opinions of the original authors, and do not necessarily reflect those of Qualcomm Incorporated or its subsidiaries ("Qualcomm"). Qualcomm products mentioned within this post are offered by Qualcomm Technologies, Inc. and/or its subsidiaries. The content is provided for informational purposes only and is not meant to be an endorsement or representation by Qualcomm or any other party. This site may also provide links or references to non-Qualcomm sites and resources. Qualcomm makes no representations, warranties, or other commitments whatsoever about any non-Qualcomm sites or third-party resources that may be referenced, accessible from, or linked to this site.