Why Does Elder Research Need a Chief Scientist Committee?

Five years ago, Elder Research formed a Chief Scientist Committee (CSC), inspired by the contributions of former Chief Scientist Dr. Andrew Fast. Initially, the CSC handled the company-wide responsibilities of: vetting and hosting weekly tech talks, managing blog posts, and defining the framework for our technical job family. But we realized a deeper need.

Elder Research’s analytic expertise is increasingly distributed among many people, not concentrated in a single “unicorn” expert. The fields of Data Science, AI, and Information Strategy—our areas of greatest strength—are rapidly expanding and it is increasingly challenging to maintain breadth and depth of capability.

The CSC immediately became a valuable forum for technical leaders from across our business units to regularly gather and deliberate on strategic decisions, share thinking and resources, and initiate focused initiatives to meet emerging needs.

We soon observed that collaboration results in much better decisions, though it’s more time-consuming than the direction of a single senior Scientist.

The Chief Scientist Committee’s mission is to serve the success of Elder Research by keeping us at the forefront of data science technology and methodology. This helps us to efficiently provide value to our customers and to serve as a trusted advisor in the broader field of data science.

As “chief scientists” it is natural to play up the importance of methodology and play down the contribution of technology, but these are complementary and amplify each other’s impact.

The rise of GPU computing for deep neural networks provides a good example: The facility of deep convolutional networks for images and recurrent (or attention-based) networks for sequences gives us amazing tools for working with such data—and these are made possible by the advent of GPU (and now TPU) computing and frameworks that make it more efficient to build and deploy these models for clients.

A committee with individuals from varied business units and different interests helps us keep up with both kinds of developments.

Day to day, our CSC members work towards this joint technical-methodological mission by contributing to the wider data science community, mentoring technical staff, and directly supporting our technical work.

Participate in the broader data science community

Participating in the broader community is essential for us to thrive as a company. Practically, the field of data science­ (along with data engineering and MLOps) is evolving so rapidly that new and useful tools are released seemingly every month.

We have to keep abreast of the field and bring new ideas from elsewhere back into the company. Reputationally, we aim to be trusted advisors in the industry. We want to contribute back to the broader technical community by, for instance, attending and presenting at conferences; writing blog posts, journal articles, and books; and contributing to open-source software.

Lead our technical staff

The industry isn’t the only thing growing—Elder Research is, too! We know how quickly it becomes impractical for a small set of technical ‘super-experts’ to be deeply involved in every client project. Instead, we need our whole technical team to learn and grow, so that more and more staff can expertly take on the challenges that our clients bring. Training and mentoring are crucial.

Therefore, CSC members mentor teams and individuals across the company. Some hold regular “office hours” for drop-in troubleshooting, advice, or brainstorming; others regularly meet one-on-one with technical staff in their business unit; and many direct internal project reviews, which provide timely expert feedback for projects from an ‘outside’ perspective. CSC members also support larger technical and professional development forums including weekly tech talks, fireside chats, fun competitions, and internal training courses.

Share technical direction

Lastly, the CSC has proven to be a useful administrative “nerve center” for our company’s technical work. Because we bring together leaders from across business units, we can share information, coordinate resources, develop and lead technical initiatives, and work together to solve problems. Beyond the impact of regular information sharing on the CSC’s inward and outward-facing missions, the CSC provides a way to share “shortcuts” that allow techniques, methods, and practical know-how to spread more quickly across the company. This breaks up “information silos,” where staff in one business unit might be facing a ‘novel’ problem that, in fact, another business unit sees regularly.

Functionally the CSC members are appointed from across the company and the leader is named by the CEO. To foster connection across Elder Research’s offices, business units, and Senior Leadership team, the committee includes at least one member of senior leadership and of each office and business unit. In practice there are about a half-dozen members active at a time, and membership rotates regularly to insure a diverse set of technical talent and experience.

Our CEO, Gerhard Pilcher, has described the CSC as a “capabilities multiplier.”  Several key initiatives have proven valuable to us so far, spearheaded by our technical leadership.

GPUs and deep neural networks

Recognizing the rising need for high-performance graphical computing to train deep neural networks and other AI applications, the CSC invested significantly in a large multi-core server with four powerful GPUs. These networks can be trained on millions of related variables — a computing challenge akin to gaming applications with millions of pixels being accurately rendered at high speed. GPU-based machines can train industrial-strength networks orders of magnitude faster than CPU-based computers, and more “traditional” machine-learning applications are increasingly being adapted to GPU computing as well. Our GPU infrastructure has served many client projects in addition to being a valuable testbed for learning and practice.

The Elder Research Vault

In a large-scale knowledge management initiative, CSC members built the Vault—now our primary internal gateway to all work done by the company. Built atop the Redmine platform, the Vault serves as a central reference for all work projects and their related technologies, built out as a knowledge graph connecting related entities.

The Vault links to key information platforms and conforms to our information security policies, securely handling our diverse set of client relationships while serving our technical, marketing, and management personnel.

Tech Talks

The CSC streamlined the Elder Research Tech Talk process and hosts company-wide talks from outside or inside experts every week. The speaker guidelines are flexible enough to serve a wide range of subject matter and presentation styles.

Internal Project Reviews

Internal Project Reviews were established to support clients’ projects by offering perspectives and expertise from seasoned technical leaders outside of the project team. These reviews explore risks and opportunities related to key project deliverables, and their benefits have proven to be bi-directional between the reviewers and the project team. Formal constructive critiques help us learn continuously and bring new information and best practices into the broad Elder Research technical community.

The fields of Data Science, AI, and Information Strategy are exciting and dynamic, and our talented group of practitioners are drawn from a wide variety of backgrounds, educational programs, and job experiences. Our clients are continuing to deepen their adoption of analytic solutions throughout their organizations, and they need access to the best expertise to stay relevant and competitive. Elder Research’s Chief Scientist Committee helps us grow our capabilities and expertise and adapt to the constant changes in the industry, a diverse and changing workforce, and the evolving needs of our clients.