Building Technical Capacity to
Fight Housing Discrimination

What it might take for the Department of Housing and Urban Development (HUD) to leverage artificial intelligence (AI) to combat housing discrimination

The Tech Talent Project  |  May 2022

Executive Summary

The Tech Talent Project has seen interest in recruiting artificial intelligence (AI) and machine learning (ML) expertise for government. The interest is clear: AI and ML can quickly identify trends and patterns, potentially easing stress in understaffed and resource-tight agencies. However, using AI and ML can lead to significant problems if unchecked. Depending on the way these technologies are set up, they can show bias and make faulty judgments in a way that can significantly impact a person’s life — and then repeat those biases over and over on a massive scale. Government plays a critical role in not only introducing these technologies responsibly, but also overseeing their use to ensure groups are not systematically being discriminated against. 

In order to better understand what resource-tight agencies need to know before introducing technologies like AI, we examined one potential AI business case in a federal department that is often in the spotlight and just as often under-resourced: HUD’s Office of Fair Housing and Equal Opportunity (FHEO). 

FHEO is a regulatory agency under the Department of Housing and Urban Development (HUD) with the mission of enforcing fair housing laws, and it plays a critical role in overseeing how these emerging technologies can impact certain groups in fair housing. For example, when lenders and landlords use AI to systematically deny loans to a particular group or reject a particular type of tenant, FHEO has a responsibility to identify and oversee where algorithm bias is present.  

In addition FHEO faces a number of existing hurdles:

  • Housing discrimination cases in the U.S. are significantly underreported. [1][2]

  • There is a relatively high barrier to filing complaints given current submission methods.

  • HUD’s ability to review and address (conciliate) all received cases is limited by the number of people at the department and a largely manual complaints review process.

  • HUD’s electronic records are not always accurate or complete, according to a 2008 audit by the Office of Inspector General. (The office completed a review of IT acquisition processes in November 2021 and announced a new audit in March 2021, with findings pending.)

  • Current audit findings on IT acquisition from November 2021 highlight that current IT systems depend heavily on contractors, and their acquisition capacity is a key potential risk within HUD’s IT environment. [3]

  • COVID-19 has made FHEO’s mission and work even more critical, as the pandemic has spurred a “looming eviction and foreclosure crisis,” according to the National Fair Housing Association — a trend that disproportionately impacts communities of color.[4]

Previous audits support that many, if not all, of the mission challenges HUD faces are impacted by its staffing issues. Despite initial steps in the right direction — including the appointment of an experienced chief information officer (CIO) in 2021 — HUD and FHEO still face critical challenges. 

We looked at FHEO and HUD because complaints help measure the problem, so it’s important to calculate accurately and understand the breadth and depth in-house. To assess this we:

  • Review the current technical and data capacity;

  • Assess the potential for AI to meet the agency’s ability to achieve its mission goals; and

  • Recommend the technical leadership needed to move the agency closer to leveraging AI.

This case study can inform other government agencies, especially those whose workload demands greater resources. Agencies can learn what is needed to use AI and ML responsibly and begin identifying and mitigating AI that shows systemic bias.

Findings

Before HUD can advance to sophisticated AI and ML, the agency needs support and resources to strengthen its capacity to process data related to and in support of its mission. As algorithmic bias is a critical concern anytime AI or ML is introduced, FHEO will need to understand and govern these technologies for discrimination as a critical part of its mission.

HUD’s current technical and data capacity requires more fundamental improvements in order to design or use sophisticated technical tools to meet its mission. These improvements are dependent on time, focus and resources. 

FHEO as part of HUD has the potential to more effectively triage, analyze and understand fair housing complaints using data products like AI. However, it must first support more fundamental needs to build capacity and modernize its delivery.

Limitations

We edited the case for brevity and clarity, and as such, the facts and context presented within this case may not fully encompass the department’s circumstances. 

We included details most relevant to talent-related questions and added considerations that help build the foundation for leveraging data and data products more effectively, including AI. All information included in this case is publicly available.

Note: In the course of completing this case study, we uncovered several other high-potential focus areas related to AI for FHEO. These areas include the evaluation of algorithms leveraged by housing agencies or private entities for bias (e.g., in advertising properties or credit products, tenant screening, mortgage lending applications), though we do not cover those in depth in this report.

Our Framework

We adapted the “Data Science Hierarchy of Needs[5] pyramid as an anchoring framework. Designed after Maslow’s famous hierarchy of needs, the data science hierarchy, first shared by Monica Rogati, provides an instructive tool for evaluating an agency’s data maturity. Our team added the dark blue section at the bottom of the pyramid, which focuses on supporting foundational needs. As the hierarchy suggests, all organizations need to start with a strong foundation of tech capacity, data governance and social consideration, and progress through several more layers before implementing AI and deep learning. In government, the real world impacts of data governance and social consideration are profound. Algorithmic bias in AI and ML is a fundamental concern and one that an agency like FHEO would need to understand and address as a core part of its mission.

Data Science Hierarchy of Needs by Monica Rogati

Adapted version of AI hierarchy of needs by Monica Rogati, June 12, 2017

Background

The Office of Fair Housing and Equal Opportunity (FHEO) is a division of HUD, whose mission is to “eliminate housing discrimination, promote economic opportunity, and achieve diverse, inclusive communities by leading the nation in the enforcement, administration, development, and public understanding of federal fair housing policies and laws.”

According to FHEO’s fiscal 2020 annual report, FHEO had:

  • 532 full-time employees

  • 10 regional offices

  • 54 field offices

  • $72.8 million annual budget

FHEO enforces the Fair Housing Act by investigating complaints of housing discrimination, assisted by field offices and state and local agencies in the Fair Housing Assistance Program (FHAP), which assist HUD in its enforcement efforts. The Fair Housing Act prohibits discrimination based on race, color, national origin, religion, sex, disability and familial status in the sale, rental, financing or terms, conditions and privileges of a dwelling, and in other housing-related transactions. 

FHEO’s staff do critical work and face several challenges including:

  • Housing discrimination cases in the U.S. are significantly underreported. It is estimated that over 4 million cases of housing discimination occur in the United States every year — and only a fraction of those cases are reported. [6] [7]

  • There is a relatively high barrier to filing complaints given current submission methods. The department has outdated technology tools, low accessibility and poor user experience for complainants. This limits the organization’s ability to fight housing discrimination for those who need support most.

  • FHEO’s ability to review and address (conciliate) all received cases is limited by the number of people at the department and a largely manual complaints review process. For each of the past three years, the agency received between 7,000 and 8,000 complaints — significantly fewer than the 4 million estimated cases in the U.S. annually. Current processes place a high degree of importance on agency staff manually receiving, reviewing and processing complaints on a case-by-case basis. There are few effective mechanisms for collecting, triaging or analyzing consistent, rigorous complaint data. It is simply not set up to scale.

In spite of FHEO’s crucial role in law enforcement and increasing pressure to perform, the agency has struggled and needs support to scale the complaints review and dispute process.

In previous evaluations and audits, both the Government Accountability Office (GAO) and the department’s Office of Inspector General (OIG) have repeatedly identified deficiencies in FHEO’s process, leading to issues with the agency’s timeliness, documentation and data integrity. [8] These challenges are aggravated by workforce skill gaps, high rates of turnover and attrition, poor delivery of technical systems, and lack of support and coordination across the department. 

In response to these assessments, HUD plans to prioritize the following in fiscal 2021:

    1. Advancing talent and management performance, and 
    2. Improving the management and oversight of information technology at the department. [9]


These priorities provide an opportunity for FHEO (and other program offices) to build meaningful technical capacity to address existing challenges and increase its ability to protect everyone against housing discrimination.

Key Technology Opportunities for FHEO: 2022-2023

The Fair Housing Act requires the federal government to take “affirmative” steps to address the ongoing harms of discrimination, segregation and exclusion. However, based on both internal and external evaluations, HUD has consistently fallen short of performance requirements and not been able to proactively enforce laws and conduct protection activities. HUD’s broad, critical mandate presents tremendous opportunities for FHEO to protect as many individuals as possible from housing discrimination. If FHEO has the time and resources to build the operational and technical capacity to deliver on its promise, the agency could better measure, understand and address the historic patterns of housing injustice that have disproportionately affected communities of color and those with disabilities.

Build leadership in tech and modern data infrastructure to foster proactive innovation at FHEO.

To approach their mission in innovative ways, FHEO can:

1. Appoint effective technical leadership and give them the authority to manage, execute and coordinate technical capabilities.

As of December 2021, HUD was still working to fill its chief data officer (CDO) role, reporting through the Policy Development and Research Office. Positions like the CDO and related roles are key to building technical capacity, effective data strategy and ultimately AI implementation. 

The department today is operating with less than half of the staff it had 30 years ago. [10] According to a November 2020 inspector general memo:

“Over the past 10 years, HUD’s staffing levels have declined, while its programs and responsibilities have increased. Additionally, HUD’s attrition rate outpaces its current hiring capacity, and employees often do not have the right skill sets, tools or capacity to perform the range of functions needed within HUD. Leadership gaps resulting from extended vacancies and constant turnover have contributed to poor or delayed decisions and an inability to sustain positive changes. Many, if not all, of the mission challenges HUD faces are impacted by its staffing issues.” [11]

These capacity gaps impact the agency across all functional and program areas.

HUD has a unique opportunity to appoint a cohort of strong technical and data leaders to spearhead the department’s development into a data-informed, software-competent organization. These leaders should:

  • be full-time employees embedded in HUD’s subsidiary organizations and on the department’s leadership team.

  • oversee effective, iterative IT and tech management on behalf of HUD’s many priority areas.

  • understand and have experience in developing or procuring modern digital technologies (e.g., data, cloud and agile software), and know when and how (or how not) to leverage them to advance their mission.

To effectively recruit, hire and empower senior technical talent, HUD should: 

  • Address technical skill gaps that are needed to support the mission.
    The department’s mandate often exceeds its staffing capacity, and existing staff have not been chosen for the technical skills necessary to meet the organization’s growing technical requirements. (More detail related to technical gaps is included in the following section.) [12] [13]

  • Address succession planning
    About 51% of HUD’s workforce has attained retirement eligibility status. HUD predicts that by fiscal 2022, 63% of HUD employees will be retirement eligible and nearly 50% of HUD supervisors and managers will be retirement eligible. [14] This presents both a significant challenge and an opportunity for the agency to hire talent with relevant skills, including related to IT, data and technical infrastructure.

  • Reduce time-to-hire, which at HUD exceeds already long standard government timelines.
    The industry standard for hiring technical leaders is 30 to 60 days from application to starting the role. It’s significantly longer for federal agencies.

    (As of 2018, the average time-to-hire in federal agencies was 98.3 days. [15]) According to HUD, its average time-to-hire was 150, 113 and 102 calendar days in fiscal 2017, 2018 and 2019, respectively. In fiscal 2019, only 6% of HUD’s 20 program offices were meeting OPM’s standard of hiring within 81 business days. The 81-day standard does not include the time to complete security clearance before someone can start. [16] Under current hiring practices, attracting high-quality and market-competitive talent will be more challenging for HUD.

2. Update data strategy, frameworks, infrastructure and governance (improving IT management).

As in any organization, FHEO’s technical systems need to support effective delivery. An internal report from 2021 highlighted some of the causes and risks of failing to modernize essential applications in FHEO:

“A significant number of HUD’s mission-essential applications have not been modernized, which presents multiple sources of risk. These applications are hosted on legacy information systems and mainframe platforms, which are operationally inefficient, increasingly difficult to secure and costly to maintain. Historically, HUD has failed to successfully implement multiple modernization plans and projects. Leadership changes with shifting priorities and insufficient funding pose potential risk to modernization. As a result, hundreds of millions of dollars in potential savings from modernization have not been realized, and security risks have remained.” [17]

While there has been a push to modernize several systems, FHEO also needs the leadership focus, resources and capability to meet its mission effectively today. Modernization, as a larger goal, takes time and requires leadership laser focused on building technical systems and platforms that will lead to a far better customer experience. For example, if the existing legacy systems are no longer supported and prevent the agency from meeting its mission in a timely manner, they should be replaced, while working to effectively process as many applications as possible.

Thus, HUD and FHEO need to prioritize a technical infrastructure that can effectively support data collection, processing and analysis to meet its potential. This requires the internal technical leadership, staffing and capacity to evaluate, build/buy and manage those tools. Stronger data and statistical inference tools hold the potential to support more efficient and effective case processing. 

As an initial step HUD is currently addressing some broader IT modernization challenges by retiring its two mainframe platforms and migrating 19 legacy systems to modernized cloud technologies. According to an OIG evaluation from June 2021, HUD has several other initiatives on its roadmap to improve core functional ability: [18]

“These efforts include standing up a centralized, cloud-based analytics and data visualization environment … The Office of the Chief Information Officer is also improving its network and data system defenses and lowering the time from threat detection to threat elimination.”

The roadmap also described the plan to deploy “[robotic process automation] bots to automate repetitive and rules-based business tasks with accuracy and virtually no human input required.” [19]

FHEO needs to strengthen technical infrastructure and build human capacity in order to leverage analytic technologies, including AI/ML. Over time, these may show promise for more efficient complaint processing, complainant response, case evaluation and analysis of housing discrimination cases. 

But focusing on effective technical infrastructure will not be enough. For example, if the agency uses bots to automate existing bad processes that don’t focus on users or customer experience, it will have automated bad processes and systems that are even harder to change. HUD will need a clear understanding of the customer experience and outcomes desired from these systems, such as radically increasing FHEO’s ability to discover and respond to complaints.

3. Upgrade an outdated and capacity-limited complaints process.

There are three areas for technological improvements in the current fair housing complaint process.

1) Supporting greater complaint volume

FHEO completed between 7,500 and 8,000 investigations per year between 2017 and 2020. [20] [21] [22]

Decades of private and publicly funded research on fair housing in the United States has found there are likely exponentially more cases of housing discimination every year than are formally reported or resolved. [23] According to the National Fair Housing Alliance (NFHA), an independent national civil rights organization, private fair housing organizations regularly process close to three times the number of complaints (73.45%) processed by state, local and federal government agencies combined.[24] In addition, cases reported to FHEO often reflect a limited subset of reasons for discrimination, as the burden of evidence is higher for certain forms of discrimination in the complaint process (e.g., discrimination based on race or gender often makes reporting challenging). NFHA has stated that, “complaints alleging discrimination because of disability continue to account for the largest number of complaints … Discrimination based on disability is usually obvious, making it easier to detect and more practical to file a complaint.” [25]

While there are several ways for individuals to submit a fair housing complaint to FHEO, the process for submitting a complaint is difficult because of website session timeouts, lack of content structure, lack of clear guidance, no friendly mobile version, and several user interfaces that confuse the person trying to file a complaint. [26] This is not the only challenge someone seeking fair housing faces. People can have real fears about engaging with government agencies when their housing is at stake, particularly those who are living in public or rent-assisted housing. A difficult user experience can exacerbate those concerns.

Currently individuals may:

  • File a complaint with FHEO online (in English or Spanish);

  • Download a complaint form and email it to their local FHEO office at an email address listed in a separate directory (available in 10 different language options);

  • Speak with an FHEO intake specialist by calling one of two 1-800 phone numbers, or by calling a regional FHEO office listed in a directory on the FHEO website;

  • Mail a completed intake form to their regional FHEO office at an address on a directory list.

Each case is assigned to one or more individual investigators after submission. [27] Individual investigators rely on the limited information they are given to proceed with next steps. These next steps might include requesting more information. The information they request might differ depending on the investigator or the local procedures. Such protocols make it challenging for the agency to collect consistent data fields for each case, especially when cases are submitted via so many mediums and there is no protocol by which to transcribe data into machine-readable or structured formats that can then be leveraged for agency or department-wide analysis. 

Staff are legally required to finish all adjudications that they start, so it’s important for protocols and metrics to incentivize a practice of triaging submissions and tracking each lead.

2) Supporting staff to create a stronger customer experience and journey

FHEO can make the customer journey for people who may have experienced housing discrimination — and the agency staff supporting those customers — more positive and effective. The current online complaint submission process is tedious, and requires that users navigate from the HUD or FHEO agency website to HUD’s Form 903 Online Complaint, [28] a submission form with long-form text inputs. The submission form is designed to take in limited information about cases, after which point investigators are required to reach out to each individual submitter to manually ask more questions. These submissions are recorded inconsistently. This also creates a lot of work for the adjudicators.

According to the agency, FHEO manually handles 1,300 yearly complaints (e.g., in an interview by phone). With data showing that 4 million instances of housing discrimination happen annually, the small number of formal complaints suggests there are clear barriers for those who have experienced housing discrimination to formally file a complaint with FHEO and likely barriers for FHEO staff who are supporting them. Moreover, a recent Government Accountability Office report noted that intake forms aren’t always fully completed and complaints take an average of 200 days to complete. [29] The report states that:

“the typical time to complete an investigation in 1996 through 2003 was more than 200 days, with some investigations taking much longer. However, a lack of data makes it impossible to assess the full length and outcomes of fair housing enforcement activities.”

To address some of the bottlenecks above, FHEO can transform its workflow to prioritize customer experience and outcomes by conducting user research to define and analyze the customer journey. This may also help:

  • Improve case management and litigation

  • Foster more consistent legal evaluations

  • Create more efficient and accurate processing of complaints.

3) Leveraging data to drive better fair housing insights

The department will be able to better understand fair housing issues nationwide if it focuses on customers, builds consistent and effective data collection, and implements real-time analytics. Focusing on customer outcomes might drive FHEO to identify the universe of potential complaints, including tracking incoming calls. Conducting a discovery sprint to determine a limited number of technical, process and data improvements would increase the organization’s capacity to meet its mission. HUD OCIO currently supports automating routine, repetitive, rules-based tasks. This can indeed free up valuable staff time for other more intensive or customer-facing roles, which would enhance capacity across the agency. The cloud-based analytics platform (referred to as EAP at HUD) is a promising first step toward more transparent, timely and detailed data reporting about housing discimination, beyond what is annually reported to the public. This type of data can be leveraged by fair housing agencies, and also by advocates and academic researchers, to advance housing equity.

A Technical Talent Strategy for FHEO

With incoming funding from the infrastructure bill adding up to $150 billion, [30] HUD has a rare opportunity to strengthen its core programs — including FHEO — with the technical and data capacity and leadership to radically improve how it delivers services, regulates housing, provides housing assistance and more. The pyramid of needs we referenced at the beginning is the first step in maturing toward AI technology in complex organizations.

The AI Hierarchy of Needs Pyramid with an indicator titled "Most agencies are here" pointing to the segment "Support foundational needs. Assess current team structures to meet mission goals."

Focus on the Foundation: Hire and Support Current Staff

1. Conduct a discovery sprint in FHEO to determine the current status of systems, processes and tech, and to identify the skill sets required to hire.

The U.S. Digital Service (USDS) has leveraged “discovery sprints” in government for years to gain a better understanding of an organizational landscape prior to building digital products. USDS offers a useful discovery sprint guide, which describes all aspects of a sprint. According to USDS: “Discovery sprints are a useful method to quickly build a common understanding of the status of a complex organization, system or service. They create paths toward solutions by identifying specific, actionable next steps for the people at the organization who will carry that work forward.”

2. Build and support modern technical talent and leadership at HUD, and in FHEO specifically.

In September 2020, the Tech Talent Project published Memos for a Tech Transition after bringing together more than 80 agency leaders and technical experts to outline the technology capacity, leadership and opportunities for responding to COVID-19. [31] The following takeaway was at the top of the recommendation list:

Agencies need leaders with modern technical expertise from day one. Appoint modern, tech-savvy leaders [32]into key leadership roles, especially procurement and operations, and appoint leaders with significant modern technical expertise [33] into key tech leadership roles.

In order to support the data needs that lay the foundation for AI and ML efforts, HUD needs to ensure that the current teams are coordinated, supported, empowered and led by modern technical leaders who can help strengthen customer-focused processes and use technology to effectively deliver services. Given the complexity and depth of data sources and programs within HUD, the agency needs modern technical leaders to help identify challenges early, coordinate opportunities often, and build and scale the needed data foundational elements. These leaders will enable HUD and its program offices, including FHEO, to meet their mission of creating strong, sustainable, inclusive communities and quality affordable homes for all. 

In Memos for a Tech Transition, we recommend that agencies:

  • Pair nontechnical leadership with modern technical advisors and bring them to the table.

  • Prioritize strong data leadership, especially in agencies considering AI and machine learning.

  • Grow the modern tech pipeline at all levels to enable an agency to effectively use and buy tech.

  • Invest time and resources to upskill America’s existing workforce. [34]

Additional strategies to build effective technical teams within FHEO include:

  • Review and structure existing teams to better support and improve service delivery.

  • Engage effective technical leaders and give them the ability to make technical and data decisions for the organization, in partnership with the chief information officer (CIO) and chief data officer (CDO).

  • Hire a CDO who can partner with the CIO and Policy Development and Research team to lead transformative data efforts on the operational side.

  • Bring HUD’s CIO and CDO teams to the table to transform FHEO, and ideally bring them to the table to support HUD’s policy implementations.

Ultimately, agencies should prioritize hiring effective technical leaders and teams. By using basic HR tools for engaging scarce technical talent — like using fellowship and hiring authorities to their full extent and tapping into the Subject Matter Expert Qualification Assessment program — HUD and components like FHEO can quickly and effectively build a staff with modern technical expertise.

3. Use best-in-class data and tech procurement approaches.

Over the past eight years, the TechFAR and the Defense Innovation Board (DIB) Software report identified several best practices and approaches for procuring technology. The DIB identified three themes that are highly relevant to HUD. These three themes are summarized below:

  • Reduce speed and cycle time for procurement activities. “Speed and cycle time are the most important metrics for managing software. To maintain advantage … [organizations need] to procure, deploy and update software that works for its users at the speed of mission need.” [35]

  • Hire technical talent to support procurement activities. Those making purchasing decisions on behalf of government organizations should be familiar with the technical challenges faced by the agency, and knowledgeable about the program issue a given software tool is meant to solve. This is especially important in the case of AI technologies, where some tools may be moving beyond data organization or analytics to provide recommendations or make decisions about outcomes, which makes program context especially essential.

  • Build capacity for continual monitoring and transformation. “Software is an enduring capability” and teams must support and continually improve it throughout its life cycle. Agencies will need to build the muscles “to enable effective delivery and oversight of multiple types of software-enabled systems, at scale, and at relevant speed.” [36]

Strengthening HUD and FHEO’s capacity to hire and support technical leaders and procure technology effectively will begin to enable the agency to better collect, move and store data. These leaders can, with the help of the CIO and CDO, begin moving toward automated processes that actually help consumers and reduce staff time.

Acknowledgements

The individuals listed below generously offered their input on AI technology in agencies and lent their time to this project in myriad ways. We deeply appreciate their time and counsel. The contents of this report do not necessarily reflect the views of those with whom we engaged, and the views of participating federal officials do not necessarily reflect positions or policies of the federal government or its agencies.

References

References
1 https://nationalfairhousing.org/wp-content/uploads/2017/04/TRENDS-REPORT-4-19-17-FINAL-2.pdf
2 https://drive.google.com/file/d/1-qkD1FQj8GjOT2UdF4buBaJ74or56_qn/view
3 https://www.hudoig.gov/sites/default/files/2021-11/2020-OE-0004.pdf
4 https://drive.google.com/file/d/1-qkD1FQj8GjOT2UdF4buBaJ74or56_qn/view
5 AI hierarchy of needs, by Monica Rogati, June 12, 2017.
6 https://nationalfairhousing.org/wp-content/uploads/2017/04/TRENDS-REPORT-4-19-17-FINAL-2.pdf
7 https://drive.google.com/file/d/1-qkD1FQj8GjOT2UdF4buBaJ74or56_qn/view
8 https://www.hudoig.gov/sites/default/files/documents/IED-07-001.pdf
9 According to a memo summarizing management and performance challenges for the agency issued by the Office of Inspector General (OIG) in November 2020, “human resource management challenges” and the “management and oversight of information technology” are among top priorities for HUD in fiscal 2021. https://www.hudoig.gov/sites/default/files/2021-06/2021-OE-0003.pdf
10 https://www.politico.com/news/2021/03/10/marcia-fudge-hud-474945
11 https://www.hudoig.gov/sites/default/files/2021-06/2021-OE-0003.pdf
12 https://www.hudoig.gov/sites/default/files/2020-12/TMC%202021.pdf
13 https://www.hudoig.gov/sites/default/files/2021-08/2020-OE-0002.pdf
14 https://www.gao.gov/assets/gao-13-282-highlights.pdf
15 https://www.opm.gov/news/releases/2020/02/opm-issues-updated-time-to-hire-guidance/
16 https://www.hudoig.gov/sites/default/files/2020-12/TMC%202021.pdf, p. 45
17 https://www.hudoig.gov/reports-publications/report/hud-it-modernization-roadmap-evaluation-report; https://www.hudoig.gov/sites/default/files/2021-06/2021-OE-0003.pdf
18  https://www.hudoig.gov/sites/default/files/2021-06/2021-OE-0003.pdf, p.19
19 https://www.hudoig.gov/sites/default/files/2021-06/2021-OE-0003.pdf, page 20
20 https://www.hud.gov/sites/dfiles/FHEO/documents/FHEO-Annual-Report-FY2020.pdf, page 14
21 https://www.hud.gov/sites/dfiles/FHEO/documents/21FHEO%20Report%20Final%201-15%20-%20Web%20Version.pdf, page 9
22 https://www.hud.gov/sites/dfiles/FHEO/images/FHEO_Annual_Report_2017-508c.pdf
23 https://www.huduser.gov/portal/publications/housingdiscriminationreports.html
24 https://drive.google.com/file/d/1-qkD1FQj8GjOT2UdF4buBaJ74or56_qn/view
25 https://drive.google.com/file/d/1-qkD1FQj8GjOT2UdF4buBaJ74or56_qn/view
26 https://www.hud.gov/program_offices/fair_housing_equal_opp/online-complaint#_How_to_File
27 https://www.hud.gov/program_offices/fair_housing_equal_opp/complaint-process
28 https://portalapps.hud.gov/FHEO903/Form903/Form903Start.action
29 https://www.gao.gov/assets/gao-04-463.pdf
30 https://www.hud.gov/sites/dfiles/PA/documents/HUD-AJP-Housing-Factsheet.pdf
31 https://bit.ly/32zaM8k
32 From Memos: “Tech-savvy leaders have responsibilities that extend beyond technology systems, but which ‘require an understanding of modern technology and the ability to hire, retain and effectively use technologists for policy, digital service and innovation.’ See also: Partnership for Public Service and Tech Talent Project, “Tech Talent for 21st Century Government,” April 16, 2020. Retrieved from https://bit.ly/32zaM8k”
33 From Memos: “Someone with significant modern technical expertise should have experience and a proven track record using a consistently evolving, iterative approach to technology to deliver effective, continually improving services rather than simply completing projects, and using the best-in-class technologies that support this approach.”
34 https://bit.ly/32zaM8k
35 Software Is Never Done: Refactoring the Acquisition Code for Competitive Advantage. Defense Innovation Board. May 3, 2019 https://media.defense.gov/2019/Apr/30/2002124828/-1/-1/0/SOFTWAREISNEVERDONE_REFACTORINGTHEACQUISITIONCODEFORCOMPETITIVEADVANTAGE_FINAL.SWAP.REPORT.PDF, p. I bottom
36 Software Is Never Done: Refactoring the Acquisition Code for Competitive Advantage. Defense Innovation Board. May 3, 2019 https://media.defense.gov/2019/Apr/30/2002124828/-1/-1/0/SOFTWAREISNEVERDONE_REFACTORINGTHEACQUISITIONCODEFORCOMPETITIVEADVANTAGE_FINAL.SWAP.REPORT.PDF, p. ii top