Interview: Identifying talent at Nasa using data science

Analysing workflows and identifying openings

Meza says one of the primary projects he is leading in the area of human capital is the roll-out and implementation of what Nasa calls the enterprise data platform, which is a group of tools and an associated framework that aims to modernise the organisation’s analytical workflow. He says this work stretches from everything to the agency’s data sources all the way to the presentation layer it uses.

“That work involves updating and upgrading our data warehouse and our data lake-type capabilities, enhancing our analytical workflow, and utilising cloud technologies, as well as other mechanisms,” says Meza.

“It’s all about looking at how we catalogue our data, how we track it, how we create metadata and data dictionaries, and then on to the visualisation aspect – and there are various ways of doing that, too.”

As part of that work, Meza is looking at how the human capital team enhances its analytical workflows and data models. That exploration exercise involves investigating how the organisation might use machine-learning technology to ensure it keeps track of its data models, and how these models might be refined operationally.

Another major element of Meza’s work is around the continuing support and development of Nasa’s talent marketplace, which is the agency’s internal job-listing and candidate-selection platform. This marketplace gives employees access to a range of internal career development opportunities at their local centres and across the wider organisation.

This platform forms part of the agency’s Future of Work initiative, which is an investigation of the roles and skills that might be required during the next 60 years at home and in space. Meza says his team continues to refine how data is used in the marketplace as part of this long-term initiative.

“We’re doing a lot of different assessments,” he says. “We’re looking at evaluating and validating some of the models that we’ve already developed to try to see if we correctly identified the right types of knowledge and skills that are within the workforce.

“We’re doing some training needs assessments and literacy assessments. Late last year, we started a digital transformation effort, which is still going go on through this year. That work remains in development and we’re still trying to work through some of those things.”

Implementing graph technology to show relationships

What is already clear, says Meza, is that data talent is spread across Nasa, some of which is not easily identified or categorised due to the wide range of work taking place at the space agency. Help comes in the form of another of Meza’s key initiatives – a talent-mapping database to identify the data skills required for all kinds of projects.

The talent-mapping database is still being developed, but it uses Neo4j technology to build a knowledge graph. Meza first started using Neo4j’s knowledge graph more than a decade ago. In the case of human capital at the space agency, the graph is designed to show the complex and varied relationships between people, skills and projects.

As a starting point, Meza’s team focused on creating an occupational taxonomy, which analysed the key components of a role from an employee, training and project perspective. To help build this taxonomy, they made use of a database from the US Department of Labor called O*NET, which has descriptions and skillsets for hundreds of occupations.

Capturing those components allowed his team to build a model of the skills associated to different roles at Nasa – and to start identifying people with skills who could fill those occupations. The model also highlights the kinds of ability that other individuals in Nasa might need to complete tasks in each occupation successfully.

Meza and his team will then get employees across the agency to validate the skills and tasks associated to each occupation, and they will use this feedback to train the model. They will then develop an end-user application and create an interface. Hopefully, by the end of this year, people in Nasa will be using the system to search for talent and potential job opportunities.

Managers will be able to use the talent-mapping database to track skills gaps and boost training in hard-to-fill areas, while employees will be able to use the technology to discover how to upskill if they want to move into new areas of work.

“Within our talent marketplace, we offer opportunities to employees across Nasa to do different types of work,” says Meza. “Having a work role that is connected to the system will allow us to hopefully find individuals that closely match what we’re looking for. Conversely, an employee who really wants to start doing a particular role can find out what kind of opportunities are out there and think about how to reach that destination.”

Helping to shape next-generational capability

Meza says that implementing all these data-led technologies is just one element of the puzzle. To really help Nasa find and train the next-generation talent it needs, managers and employees will have to get used to making the most of the technology that his team is implementing. Meza recognises that process requires a shift in mindset.

“It involves a lot of communication, and sometimes a lot of grovelling and begging,” he says. “I say that tongue-in-cheek, but it’s hard to move an organisation – not just Nasa, but any organisation – that has an embedded system and to try and get them to do something different. A lot of the time, that means working in parallel to begin with, which means added costs before you can completely switch over.”

However, Meza believes this joined-up approach will help everyone in the organisation to begin to see how data and technology can be applied in combination to help Nasa overcome some of its most intractable skills challenges. As the technology proves its worth, he expects the agency to begin to think about how graph systems might be applied in other areas.

For now, Meza is keen to continue developing innovative ways to help Nasa fill its most significant skills gaps. When he thinks about the continuing requirement for talent in the agency, he believes the data scientist of the future will be part of a cohesive and collaborative group – and the technology that his team is developing right now will help point skilled individuals in the right direction.

“As we start to develop our infrastructure and our capabilities to have model libraries and to share code more easily, I think you’ll see data scientists that can easily pick up different domains based on these models and identify how they might be able to use these tools within their own domains,” he says. “So I think shareability and the use of models is definitely something that’s going to increase.”