Preparing for future missions to the Moon & Mars with dynamic workforce scenario planning
Company
National Aeronautics and Space Administration (NASA) is an independent agency of the U.S. federal government responsible for the civil space program, aeronautics research, and space research.
Size:
18,000 Civil Servants
Industry:
Public Sector
Partner Integrations:
AWS, Amazon Bedrock
Technology used:
Posit Team (Posit Workbench, Posit Connect, Posit Package Manager), Dash, Shiny, Quarto, Vetiver, Shiny Assistant
Apps/products/solutions delivered:
Demand scenario forecaster, visualizations, models, Tableau integrations, and many more
"The combination of AWS and Posit has transformed our analytics organization from a traditional reporting function into an AI-powered innovation engine."
David Meza, Head of Analytics - OCHCO and Branch Chief - People Analytics at NASA
The Challenge: Improving the speed of iteration & response for complex questions
NASA’s people analytics team sought to improve the speed of iteration and response due to frequent, immediate requests from Congress and high-level stakeholders.
While BI tools (like Tableau) provide them with static dashboards and descriptive analytics (e.g., “How many civil servants currently work at NASA?”), they also needed to address more advanced predictive and prescriptive analytics. This shift would enable the use of interactive applications to address complex questions like “What impact would a one-year mission delay have on the workforce?”
Although R and Python users on their People Analytics Team could build models to simulate these scenarios, the models were confined to individual laptops and required physically bringing laptops from meeting room to meeting room to present results.
To better manage risk and prepare for the future, workforce planners within individual NASA Programs needed a way to slice and dice the information themselves, especially when facing new budget proposals.
"What used to take months of infrastructure planning and model deployment now happens in days—that's the edge we've been looking for."
Rapid Application Deployment for Workforce Planning at NASA
NASA's People Analytics team now uses Posit to quickly prototype and deploy applications, often on the same day a request is made. This has led to notable improvements in performance management, organizational effectiveness, and workforce planning for the agency, which includes 18,000 civil servants across 193 active and future missions.
Given the scale of its workforce and missions, NASA requires precise planning. As one example, the People Analytics team utilizes a Demand Scenario Forecastor, a Dash application built and deployed via Posit Workbench to Posit Connect to project future full-time equivalent (FTE) personnel needs. This tool accounts for factors like unique competencies and external influences such as retirement rates, economic instability, and elections over the next decade.
The Demand Scenario Forecastor enables business stakeholders to independently conduct asynchronous what-if analyses. For instance, a workforce planner can use this tool to select projects and evaluate the predicted impact of variables like delays, increased skill set requirements, or funding adjustments on necessary FTEs.
While Dash is highlighted in this example, the team has the flexibility to deploy a wide array of data products to Posit Connect. This includes applications built with Flask, Streamlit, Shiny, Quarto, R Markdown, FastAPI, Jupyter, and over seven other frameworks.
The Future of Analytics at NASA with Posit and AWS
Building on their success, NASA’s People Analytics team continues to explore and rapidly prototype AI solutions for future projects with both Posit and AWS.
David Meza, Head of Analytics - OCHCO and Branch Chief - People Analytics at NASA shares, "The combination of AWS and Posit has transformed our analytics organization from a traditional reporting function into an AI-powered innovation engine. With Posit's unified platform running on AWS infrastructure and seamless access to Bedrock's foundation models, our teams can now prototype RAG applications in the morning and deploy production-ready analytics dashboards quickly. What used to take months of infrastructure planning and model deployment now happens in days—that's the edge we've been looking for."
By transitioning to Anthropic models via AWS Bedrock, NASA was able to save over 50% of the cost to deploy to their Human Capital group of approximately 300 people.
An example of this collaboration is the team’s project tracker, where Shiny Assistant has accelerated front-end app development. With the integration of Shiny and AWS Bedrock, leaders can use natural language queries, such as "What projects are blocked?", directly within the app, and easily create monthly reports.
While reports previously required a team to manually run SQL scripts, for instance, on every third Tuesday and compile them into a Word document for distribution, this process is now automated. Quarto documents are deployed to Posit Connect in the desired format for leadership, with an option to still export to Word documents if necessary.
This streamlined approach significantly enhances efficiency and ensures that critical information is consistently available to leadership.
Additional Resources
Industry
AWS
Use the power of Posit's data science software combined with the flexibility and scalability of AWS.
Blog
Deploy your models on Amazon SageMaker with vetiver
The vetiver R package now provides fluent support for deploying models to Amazon SageMaker.
Blog
How to use natural language data science with RStudio and Amazon SageMaker
Integrate Amazon Bedrock LLMs into RStudio on SageMaker using the gander package to enhance R workflows with natural language.