Posting Language
Title
Authorize negotiation and execution of an agreement with Valkyrie Intelligence LLC to develop a data-driven framework for identifying households at risk of experiencing homelessness in Austin, for a 13-month term beginning September 1, 2025, in an amount not to exceed $250,000. Funding: $250,000 is available in the Fiscal Year 2024-2025 Social Services Contracts Budget of the Homeless Strategy Office.
De
Lead Department
Homeless Strategy Office.
Fiscal Note
Funding in the amount of $250,000 is available in the Fiscal Year 2024-2025 Social Services Contracts Budget of the Homeless Strategy Office.
For More Information:
David Gray, Homeless Strategy Officer, 512-972-7836; Kelechukwu Anyanwu Jr, Program Manager III, 512-972-5136; Estella Kirscht, Department Executive Assistant, 512-972-4423.
Additional Backup Information:
This action will authorize the Homeless Strategy Office (HSO) to negotiate and execute an agreement with Valkyrie Intelligence LLC (Valkyrie), beginning September 1, 2025, to develop a data-driven framework for identifying households at risk of experiencing homelessness in Austin. The primary goal of this effort is to shift the City’s approach to proactive prevention through the creation of a robust data strategy and predictive modeling methodology.
During this engagement, Valkyrie will conduct exploratory data analysis, assess current and available data sources, and define the technical and ethical foundations required for predictive modeling. Key activities will include stakeholder interviews, data access and documentation review, algorithm selection, and the design of a responsible AI strategy tailored to the City’s homelessness prevention objectives.
The project will culminate in a comprehensive findings report, a model implementation document detailing the approach to risk prediction and ethical deployment, and all final code assets. There will also be a final readout to present findings and recommended next steps, which may include a second engagement focused on building the predictive model.