As Nomadik starts to explore all of our data streams and our models start to build connections we noticed that the correlation between current housing placement, eviction processes, and homelessness has risen. BASTA Austin, a tenant legal aid NPO, has recently released a report of the top properties pursuing evictions around Austin from 2023. Abelia flats, a property used for housing placement by organizations such as Front Steps, was among the properties listed. Nomadik uses a distinctive data feed to better understand the homeless crisis which involves all surrounding facets of homelessness. This data includes fire and EMS reports, police incident reports, eviction reports, local map data, and more. Our models use these reports to build out robust connections and findings on the causes of homelessness and the relapses of those who had already been placed into housing. In this specific case we explored the APD incident reports to build out more robust connections in our models and what we found is extremely interesting.

Source: @bastaaustin

Looking at APD incident reports over the first three months of 2024 we found that Abelia flats had a statistically significantly higher rate of assault incidents than the surrounding apartment complexes, around 4:1 ratio on average. For the most part, when looking at incident reports for apartment complexes the reports with the highest frequency are burglary of vehicle, but for Abelia this was not the case, in fact in that period of time there were no reported burglary of vehicle incidents. Extending the time window further shows that the rate of assault in Abelia flats is much higher than surrounding apartment complexes, so why is that? Immediately pointing to the rehousing of clients exiting homelessness as the perpetrators is not fair as we do not have the incidents mapped to individuals, nor do we have the data on which clients were housed at Abelia flats, but we can look at the correlation.

Over the past two decades many studies have discussed the impact of supportive housing and mixed traditional low income housing. This is not to be confused with Section 8 or project housing. It begins with low income apartment complexes being offered housing vouchers from various organizations, the city, and state housing programs. These housing programs offer assured payment of a rental unit whether the client stays or decides to return to the streets. The idea is to hold the unit for a year and if needed, try to fill it with the next individual in need. This presents more benefit to property management and owners as a promised revenue for however long the voucher dictates as dependable rental income. For most, accepting guaranteed income is a no brainer, why deal with tenants that may be late on rent when the government can guarantee the income stream?

Many clients continue hoarding tendencies after being housed resulting in similar living conditions to outdoor encampments.

The studies mentioned above also point out that individuals under these supportive housing programs tend to bring their lifestyles, habits, and addictions from the streets to these apartment complexes. One common habit for those who have lived on the streets for extensive periods is hoarding. It is a common survival trait from the trauma of loss and always being in need of basic necessities. These factors lead to an increase in crime and adverse living conditions for low income households that are working and paying rent with none of the assistance provided to the housed clients. Furthermore, damage ensues while property and community value declines. This is normally a result of clients with addictions, mental illness, and the visitation of others still living on the streets. While the intention from a client might be genuine to offer friends a shower or a night's sleep in a safe place, it usually ends up with a chaotic situation. The organizations housing these individuals fail to keep up with consistent daily check ups and what is known as "wrap around services" from housing case managers. The ultimate conclusion for these properties is that the value decreases, the normal revenue for rents slows down to a crawl and eventually the property shuts down. Once this occurs the property is now in the flux of a development sale, possible bankruptcy, or a sale to the city where it will be redeveloped for new upper income earning tenants.

Now what do these banks do with the properties they acquire? Generally they shut them down, evicting all of the tenants, either demo and rebuild the property or renovate it and reopen it as a “luxury” apartment complex to attract high paying tenants. The “doom loop” described in the previous paragraph plays out all too often across the United States transferring locally owned properties to banks and conglomerates. What started out as the apartment complex looking for a nice source of guaranteed income, and perhaps wanting to support their local community, turns into a financial catastrophe doing more harm than good to both the apartment complex and the surrounding community.

Without supportive programs for mental health and addiction clients are not able to thrive with the opportunities provided to them.

Now let's try to understand the situation at Abelia Flats. It is evident that the property is not necessarily at fault but more so the "Housing First" HUD mandates and the local/municipal dealings between housing nonprofits, politics, and the inability to keep case managers consistently employed. Therefore failing to provide consistent case management for the clients most in need and the lag of rental payments that also become a warning to other properties not to accept these clients from particular nonprofit organizations. This happens throughout the nation in cities with the highest homeless populations and has been the cycle for at least the past five years running side by side with the rise in homelessness. It almost seems as though the plan is to fail and keep the mindless rotation going.

What is happening is that local nonprofits are taking clients and housing them before addressing the issues that likely drove them into homelessness in the first place. These clients then continue to live as they did before without the required support of mental health care and drug rehabilitation but now are in a community of people who have not experienced homelessness nor are prepared to handle situations that arise from living next to the mentally unwell and drug addicted. This can lead to a higher rate of assault as we have seen in our data among other incidents generally not commonplace among the average population of an area. The combination of the higher potential for incidents as well as any other factors that go into determining if an apartment complex will serve an eviction points towards the reason places like Abelia have higher eviction rates than surrounding complexes, but that isn’t the whole story.

Haven for Hope is a privately funded navigation center in San Antonio, TX that places all the resources someone exiting homelessness could possibly need to make the transition as smooth as possible: healthcare, addiction treatment, ID services, and more.

The supportive housing vouchers come in 2 forms:
Permanent Supportive Housing - a program that renews annually and usually works out well for those with real disabilities, elders, and veterans that might not have gotten into a veteran housing program just yet.
Rapid Rehousing Program - a 1-2 yr program designed to help capable clients get back on their feet. The first year is guaranteed, dependent on client behavior. The second year is granted in a situation where a client is making real progress and genuinely needs a little more time with evidence of progress.
Once the program is no longer supporting the client the tenant needs to have found a way to cover the rent themselves. Without mental health care and drug rehabilitation the odds of the tenant acquiring the means to cover the rent is hampered which leads to an observed increase in evictions. For housing to be the life changing solution that “Housing First” works towards, the system needs to be reworked to include the services required to address the needs of a population that has an extremely high rate of mental illness and drug addiction. Without addressing the underlying causes the same cycle of homeless to housed to evicted to homeless will continue for large portions of the homeless population. 

If you follow us or DASH on Twitter you’ve probably seen our coverage of the camp just south and up hill of Barton Creek on the Barton Creek Greenbelt. It’s not on any of the trails, but it does exist within the boundaries of the Greenbelt itself. This camp sits on the corner of Lamar and 360 behind the old Toys R Us, it is about 1 square kilometer in size and contains an estimated 100 tons of trash and debris. This camp popped up on the city’s radar when a fire broke out in the camp on March 2nd of 2024. Since then we have worked with DASH and our own outreach team to use this camp as a case study on how camps start, grow, evolve, draw the attention of the city, and subsequently are cleaned up and where everyone goes following the closure of the camp.

The large camp at the bottom is the camp this post refers to.

Over the past two months we’ve spent a considerable amount of time in the camp documenting what is going on and how the city is addressing it. We also spent time talking to those who live there about how they got there, how things have changed, and where they plan to go. We’ve spoken to some who have been homeless in Austin for over a decade who told us about the evolution of homelessness in our city. Prior to 2017-2019 homelessness in Austin was viewed in a much more positive light, there were even famous homeless people, adored by the city, such as Leslie Cochran. The media surrounding homelessness was more positive, there was hope in the city for helping those that were not viewed as hostile vagrants, but as neighbors who simply needed help getting back on their feet.

Estimated number of people experiencing unsheltered homelessness in Austin (ECHO).

One resident of the camp told us how he used to live in an area near the Barton Creek Mall where he had an agreement with the city that he could live in the woods nearby as long as he cleaned up his own trash and whatever trash was found out in the area around him. He told us he found hundreds of used propane tanks, decades old at this point, that took him years to finally clear out. Every week he would take his trash and whatever propane tanks he had collected over the week and place them at a designated spot by the street and the city would come collect the trash. In 2020 COVID-19 hit the US, the city shut down all of the shelters and reduced outreach and city services to stop the spread, pushing everyone back onto the streets. At the same time Fentanyl was taking off in the US, contributing to a sharp rise in overdoses and replacing the previous drugs plaguing the homeless community. These were coupled with a sharp rise in the homeless population to include, as the same resident told us, people who came in and for one reason or another started trashing camps that had previously been clean by comparison.

These compounding factors have lead to the situation we see today: out of control camps effectively creating land fills in the greenbelts, a sharp rise in violence out of the community, and an animosity from the general population towards the homeless community. After the fire in March the city came in and put up notices around the camp to tell residents it was being shut down for cleanup. The city is not allowed to shut down a camp unless they have enough beds to support the entire population of the camp, in this case somewhere around 40-50 beds. While some residents went to housing others decided to pack up and leave, some that took beds only used them for showers but still remained in the camp. Those who did not take the beds did so for a variety of reasons, some were unable to go to the shelter as they did not want to give up their pets, some told us that the shelter was often times more dangerous than the streets, and some just wanted to live outside. It has been two months since the city “shut down” the camp to clean it up. There are still people living out there, the city did not even come out to clean in earnest for several weeks causing some of those that had moved out to move back into the camp.

An abandoned tent complex inside the camp.

When the city did show up they brought heavy equipment. Below is a topographic map of the camp, the camp sits on an extremely steep hill south of Barton Creek, making the use of heavy equipment an extreme challenge. They initially cleared the area directly behind the parking lot behind the Toys R Us and the apartment complex with a bulldozer collecting roughly 8.6 tons of trash and debris. The area they cleaned was about the width of a 2 lane road along the top most part of the slope around the areas mentioned previously. The city then did not return for another few weeks, where in that time new comers showed up and undid all of the work the city had previously done. The city showed back up again the week of April 29th and cleaned the area again, along with areas below the initial cleaning area where they could get the bulldozer down to.

Topographic map of the camp.

Having spent considerable time in the camp with DASH we can tell you that the areas with the worst debris fields are much farther down in the camps as is shown with red circles in the topographic map above. If you follow DASH you have seen how steep the trails are in the camp, but while the bulldozers may be able to handle steep inclines the camp is also nestled in the trees, trees close enough together to prevent a bulldozer from moving freely. The city has claimed they are unable to use heavy machinery due to endangered species in the area. To clean this camp would require hundreds of people.

A gnome mask tied to a tree in the camp.

A question we at Nomadik have is how does this camp, uphill from Barton Creek affect Barton creek itself? Does runoff from the debris, trash, and hazardous waste effect the water that feeds Barton Springs? On May 9th we set out to collect the data required to answer this question. As it turns out the creek is completely dry, suffering from a 2, almost 3 year drought. There are one or two areas of standing water from the recent rains and a stream that feeds the creek, creating a small pond from the pond just below the 650 mark on the topographic map above, but nothing more. While we did take samples from these areas that had water, none of them were in an area where the camp runoff would reach them. So we hiked up the ravine that goes into the middle of the camp to see how far down the debris had come, given that where the ravine meets the creek we saw no trash. The further up the ravine we went the more trash we found, stuck in fallen trees that had built up natural silt barriers with detritus.

Hiking Barton Creek to collect data.

While the current state of Barton Creek has staved off the potential for hazardous waste, trash, and debris from flowing down to Barton Springs and Lady Bird Lake the state of affairs doesn’t seem to be changing at a pace that may one day lead to this outcome. This is just a single camp, near a single creek in Austin. We have over 180 documented camps, with a large number of them existing in green spaces near the creeks, streams, and rivers that flow through our city. When a camp this size is closed, the residents who do not enter housing have to go somewhere else, the portion that contributed to the mess in the first place simply do the same thing somewhere else, creating a feedback loop that the city is constantly playing catch up on. Nomadik set out to use advances in data science over the last several years and methodologies previously siloed in completely separated industries to better understand the homeless problem and empower those that directly work in the space every day: first responders, non-profits, the city, and every day citizens. To do this requires far more than “introducing tech to a new industry” or building an app, it requires a sense of civic duty to get out into the city to do the ground work that will lead to Austin becoming a shining example of cities in the United States.

P.S. Do not try to take short cuts on the Barton Creek Greenbelt, while they may be short, there are also 150 foot cliffs.

Prior to starting this journey we worked in defense tech building AI/ML solutions for the US Department of Defense (DoD), primarily the Navy and the Air Force. Defense tech is an interesting space as it provides the opportunity to work on issues completely off the radar of every other industry, especially when it comes to AI/ML. For those familiar with geospatial intelligence, or commercially GIS (geographic information systems), applying location and time data to almost any data type provides a wealth of information compared to having that data in a vacuum. After years of working with camera data, satellite data, and any other kind of temporal location data you can think of, we realized that this specific way of understanding large scale systems was mostly confined to the DoD and industries that used GIS. Not only that, but with the AI/ML explosion of the past few years we rarely saw companies applying the new state of the art to actual issues everyday Americans faced. 

Understanding the location helps understand the impact to the environment surrounding it.

Defense tech was still fairly novel when we started, but now sees an increased interest in the form of American Dynamism. Outside of this space, where AI/ML is being used to further national interests, the primary use cases seem to be entertainment, finance, customer service, and software development. The products being developed are extremely ground breaking and will super charge their respective industries however none of these address the issues you and I face in our day to day lives outside of work and entertainment. As we started to explore what societal issues we wanted to tackle we realized that all of the major issues in the US coalesce within homelessness: mental health, drug addiction, crime, violence, and not to mention homelessness itself. Combined with the fact that homelessness has seen a sharp increase post-COVID the field seemed like the logical choice to apply our talents to an issue where temporal location data actually provides extremely valuable insights. 

Processed Skyfi satellite data showing a construction site caused camp movement.

Current HMIS (homeless management information system) guidelines put forth by the US Department for Housing and Urban Development (HUD) collect client data at a basic level to coordinate access to local resources like shelters, housing, etc. They also require a point in time count (PIT) every two years for local jurisdictions. These two primary sources of data on homelessness exist on the opposite ends of the spectrum, granular client level data and macro picture counts that still rely on coordinating hundreds of volunteers for one day every year or two to attempt to get an accurate overall count. There does not exist anything in between, no one is measuring the impact of closing one camp and how that leads to some amount of new camps. No one is measuring how the opening or closing of a specific resource in an area affects local populations, businesses, and green spaces. 

The data as it stands exists in a vacuum with extremely understaffed agencies tasked with collecting it. The first step to the solution, the step we are taking, is democratizing outreach. Put the power of contribution and situational awareness into everyone's hands to augment the systems already in place. Apply temporal location data to that information and turn it from a random one off observation into useful information we can combine with all of our other sources to start to understand patterns and systems. 

Out in the camps with DASH.

Once we start to use our models to pull these patterns out we can start to address the root causes of issues. Look for upstream levers we can pull that have outsized downstream effects, beginning to act rather than react. The US has a $10 billion budget for homelessness, but they have no real way to quantify the effects that amount of money has on a problem as dynamic as this. With a data layer being fed by every citizen that seeks to help we can start to understand what works, what doesn’t, and how we can better allocate these funds, the policies, and the resources. 

Building American Dynamism

We saw a gap that needed to be filled. Not just in the tech space or the social services space, but in the US as a whole. A space that exists on the other side of the American Dynamism coin where advances in manufacturing, aerospace, energy, and defense require a strong, safe, and resilient society. A gap that requires bringing together aspects and technologies from those places and applying them to industries and systems that previously had never considered the use of something like geospatial intelligence to solve their problems. To push America forward we need to pull America back onto its feet.

You can help us model this issue by participating in our Android and iOS beta. As with any problem individuals, communities, and society as a whole faces, the more data you have the better you can understand and build solutions towards solving said problem.

The United States metro areas face a seemingly insurmountable crisis - homelessness. In 2013 the US Government implemented the Housing First model as a one-size-fits-all to homelessness. It was the first in a string of overarching decisions that lacked empirical evidence and data to support the cause and effect mechanisms inherent in the complex problem. People are suffering on the streets, not receiving the support they deserve. The administration promised it would end homelessness by 2023 which fueled a guess and check mentality to solve the issue, but it led to inflating the administrative costs and tens of billions spent in public and private resources. It is 2024 and the issue is boiling over. The population has become desensitized to daylight theft, squalor on crowded sidewalks, open drug use in view of schools, violence and human behavior with similarities to a post-apocalyptic world. We can all point to a list of probable compounding factors: rising cost of living, lack of affordable housing, the opioid crisis, lack of access to mental health care, and on and on. One thing is certain: we are not utilizing the advancements in data science, systems modeling, and machine learning, to show how these factors affect homelessness. We are not aggregating disparate information sources nor are we providing a baseline for cause and effect behaviors to assist in diagnosing these issues. We are not using the tools and technology available to diagnose this major societal injustice. Nomadik is here to solve that. 

Figure 1. Estimated number of people experiencing unsheltered homelessness in Austin (ECHO).

The founders of Nomadik have spent the last few years building an unparalleled skill set in practical AI/ML frameworks, computer vision, optical modeling, and geospatial intelligence and are now applying these skills to address the nationwide issue of homelessness, starting in their own backyard - Austin, Texas. The data that surrounds this issue is fragmented, disconnected, dated, or in most cases does not exist. Nomadik wants to democratize the solution; all it takes is a smart phone. Our soon to be released Android and iOS app will make sending images, videos, and any other related information to the platform as simple as possible. Nomadik processes the incoming data and provides an up-to-date situational awareness map, from locations of camps and debris to resources available to the homeless. 

Figure 2. Estimated number of people experiencing sheltered homelessness in Austin (ECHO).

What users provide us is what’s known as point-in-time geospatial data. Contrary to popular belief, there is not one centralized machine learning network, or AI, performing the analysis seen on our website, but a multitude of small, purpose built models. This strategy enables us to use isolated and domain agnostic data to train individual models to complete small tasks, simultaneously breaking down the complexity of the bigger picture, and ensuring robustness and the isolation of sensitive data. These specialized models allow us to build an accurate picture of features such as camp boundaries, debris levels, hazard levels, and more. Given that submitted data is location tagged point-in-time information we are able to feed the state of the region over time to further models that forecast specifics such as impact of outreach and available resources or hazard impacts to local areas.

Figure 3. System capacity by year (ECHO).

We will be partnering with non-governmental organizations, public service providers, law enforcement, and anyone else who wants to get involved to drive effective and efficient resource management for the most disadvantaged in our communities. The current process is broken, lack of data and manpower makes concentrating efforts a monumental task. Austin’s budget to address homelessness is more than $500 million, funded through both federal programs, as well as local tax dollars and private donations. Despite massive funding efforts the homeless population continues to increase while resources are strained trying to keep up, especially within housing. In 2021 the estimated unsheltered homeless population was just over 2,000 people (figure 1) and the sheltered population around 550 (figure 2), with a total housing system capacity of 3,500 beds (figure 3), a person to bed ratio of around 1:1.5. In 2023 the estimated unsheltered homeless population was over 5,500 (figure 1) people and the sheltered population around 1150 (figure 2), with a total housing system capacity of 4,250 beds (figure 3), with the person to bed ratio now being around 1.7:1. In 2021 there were more than enough beds for the entire homeless population of Austin, TX, since then the population has tripled with housing resources unable to keep up. The Austin Police Department exhausts their budget directed towards homelessness at what amounts to addressing 109 individuals. The city has managed to close less than 50 of the more than 180 known camps throughout the city and the point-in-time counts continue to be an antiquated system to understand the state of the city at any given moment.

Nomadik is going to bridge these gaps, by providing the most comprehensive platform on this crisis to ensure that those doing the groundwork have the clearest picture possible when trying to make a difference. By democratizing outreach we expand the understanding of these camps past the resource constrained organizations tasked with the monumental task of helping those in the community and ensuring our cities remain safe and clean. Changes in camps, from population, to debris levels, to disposition are data points resource providers and the city need to properly respond to a situation that requires proper resource allocation. The same information is vital to those who live in the communities affected by the presence of encampments, whether that be hazardous materials leaking into the water supply, presence of criminal activity, or elevated levels of fire danger. By having a system that enables everyone to contribute and gives everyone access to the same information we can work towards better solutions to address this humanitarian crisis.

Sign up for our beta on our website to get involved.