As digital transformation continues, here’s how Industrial Data Scientists can help overcome challenges at industrial facilities.
By Ary Bressane, Director of Data Science – AIoT Solutions, AspenTech
The concept of the “Industrial Data Scientist” is a relatively new development in the industrial sector, but one that’s absolutely essential to the industry’s ability to compete, thrive, and innovate amid a rapid movement toward digital transformation. Bridging the gap between traditional data scientists and industrial domain experts, the Industrial Data Scientist’s core mission is to build comprehensive, performant, and sustainable AI and machine learning (ML) models that are fit for purpose in addressing real-world use cases.
That mission, though, is held back by a number of longstanding organizational, technical, and process challenges inherent in the industrial sector – challenges that make it more difficult for Industrial Data Scientists to do what they do best and deliver new, innovative value for their organizations.
What are the institutional challenges stifling the work and growth of Industrial Data Scientists? What solutions can plant, refinery, and other industrial facility leaders deploy to circumvent these roadblocks? And, for that matter, just who are these “Industrial Data Scientists” to begin with?
What exactly is an Industrial Data Scientist?
Unlike traditional data scientists, who hail from data science-oriented backgrounds like computer science or software engineering, Industrial Data Scientists are a new kind of data-driven domain expert that come from less typical backgrounds. One recent survey found that nearly 40% of Industrial Data Scientists actually come to the job with a chemical, industrial, or process engineering background. In other words, Industrial Data Scientists are domain experts in their unique fields first, then make the jump to data science later, a “best of both worlds” role that overlaps domain expertise with tech-savvy data science skills.
That different identity also means Industrial Data Scientists serve a different function in their organization. Traditional data scientists tend to leverage their academic experience in how they address problems, with an emphasis on long-term projects like KPIs, evaluation metrics and ML models. Industrial Data Scientists, on the other hand, focus their problem-solving efforts on more tactile, real-world problems that occur in the field. The Industrial Data Scientist’s approach is to combine their own innate domain knowledge into the data science projects they’re working on, adding a new dimension to the work. With that domain knowledge, Industrial Data Scientists drive new opportunities for innovation and value-adding tasks for the business, while also having the technical skills needed to turn data into actionable next steps.
As larger trends around the industrial sector continue to take shape – such as digital transformations and workforce shifts seeing veteran domain experts retire and replaced by younger tech-savvy workers – Industrial Data Scientists will be integral to how their organizations transform and leverage new Industrial AI applications to continue pushing the envelope. At the same time, that growth trajectory is going to have a serious ceiling on it so long as a number of institutional challenges continue to prevent Industrial Data Scientists from being able to do what they do best.
The key challenges bottlenecking Industrial Data Scientists’ full potential – and how to solve them
In order to enable Industrial Data Scientists to put their best foot forward – so that their work can drive the rest of the organization forward – industrial business and IT leaders need to be able to identify the roadblocks impeding those best efforts. These challenges include:
- Coordinating domain knowledge resources across internal experts.
- Overcoming longstanding data integration and mobility hurdles that keep data siloed in disparate sources.
- Managing data quality, integrity, and security.
- Pooling together all necessary data engineer, DevOps, IT, and integration efforts to mature proofs of concept into the productization phase.
- Determining where to deploy, iterate, and collaborate on ML code.
- Identifying the proper AI and ML environments.
- Deploying models in production. Managing the full lifecycle of machine learning models, from experimentation to production.
- Deploying and scaling additional resources as needed to deal with growing data volumes and complexity.
- Collaborating with domain experts to continuously tune, test, train, and optimize AI and ML models for achieving business goals.
- Presenting results in an actionable, visualized manner to decisionmakers.
Leaders at industrial facilities – plants, refineries, steam crackers – can better support their Industrial Data Scientists to overcome these challenges and focus on their core mission by arming them with the tools they need to succeed. In other words, investing in a robust, scalable, and secure Industrial AI infrastructure. That means a full-stack approach to:
- Abstracting complexity in data science projects.
- Democratizing the ability to fetch critical datasets.
- Facilitating AI and ML project deployments into production.
- Streamlining collaboration across domain experts and engineers.
It’s that Industrial AI infrastructure that creates an environment where Industrial Data Scientists are free to develop, train, maintain, deploy, and execute Industrial AI and ML models, without any of the usual roadblocks. In other words, an environment where Industrial Data Scientists can do what they do best: accelerate time to market, increase productivity, drive innovations, and deliver new value to their organizations.
About Ary Bessane
Ary Bressane is the Director of Data Science at AspenTech and leads the development of new products and technologies related to Data Science, Artificial Intelligence, and Internet of Things. With over 15 years of industry experience, Ary builds innovative data-driven solutions to inspire the next generation of Industrial AI products and drive business impact.