Finding the perfect candidate is difficult in normal circumstances, but in the specialized world of high performance computing (HPC), accessing experts or developing the next generation of HPC talent becomes a differentiating capability. The demand and supply of specialized talent have reached an equilibrium with competing market and geographic factors holding the strongest individuals in place, forcing creative approaches to delivering the right expertise at the right time. Additionally, there are not many “Super Computing Universities” churning out mission-ready HPC junior team members.
As a program manager, I spend as much time getting the right individuals in place as I do thinking about how the application of HPC will address my customer’s challenges. My current program supports climate research and the use of HPC to create large high-resolution predictive models in support of NOAA’s larger climate mission. How our scientific customers use HPC differs from how a defense or intelligence agency might use HPC, and that directly impacts my hiring strategy. For instance, my team needs varying degrees of expertise in climate and other natural sciences paired with IT and computer science within HPC applications. We also staff a range of commodity IT professionals who provide overarching infrastructure and administrative support, insulating the climate scientists from the technology so they can focus on research. The result is that my primary staffing goal becomes seeking the mix of talent that can best enable the science.
Accessing ExpertiseWhen filling a full-time need, I find it especially important to take advantage of opportunistic timing as talent moves through the industry. But what do you do when the “free-agent” pool is shallow? For short-term needs, I leverage the reach-back resources of Engility. Engagements such as commissioned studies and temporary surge support are the usual success stories for this approach; however, long-term partnerships and dialog with other HPC programs yield immense returns in long-term strategies and thought leadership. Investing in this concept, Engility established an HPC ENnovation Center, which helps me access pockets of HPC expertise in many specialty domains.
HPC programs are often siloed from one another and, despite how big the enterprise is, they tend to work like independent businesses. This isn’t deliberate; many programs are just saturated with work, and it’s difficult to take steps back to look at the bigger picture. Having reachback is extremely valuable, and exploiting it well differentiates you from your competitors and builds confidence from your customers. Ultimately, it’s really up to program managers to convert that capability into value for our customers.
A Good Farm SystemThe availability of capable new talent is challenging in STEM fields and is obviously critical for staffing in a HPC environment. Like a baseball team, it would be great to have nothing but experienced all-stars, but that is not fiscally feasible and does not always fit the program needs. Sometimes this means your mix is heavier on the junior end with the goal of enabling upward mobility for high performance individuals. Other times it’s correcting a top-heavy staffing mix with not enough operational support to address the finer details. Staffing mix changes do not happen in a vacuum, and your customer must be comfortable with the pendulum swing over time and have confidence in the mix that your staff provides. As the program manager I am constantly communicating with my customer, introducing young talent and balancing the risk of inexperience with the costs of seasoned veterans.
Attracting actual junior staff is another problem completely. In the best traditions of major league scouting, I need to hire for potential rather than expertise. I work with my talent team to build professional profiles to identify preferred skills and qualities that individuals must possess to do well on the program. Equally important to evaluating skillsets is an understanding of geographies and what motivates different people in different positions on your program…think Maslow.
An understanding of available free-agents, a strong farm system and a robust reach-back capability give me the tools I need to build an effective HPC team. In sports, there are going to be championship years and rebuilding years, but if I manage our talent mix to mission requirements, I am setting up our customer for success in all seasons.