Our People: Edge AI, Harm Reduction, and a Woman in a Wheelchair
- John E. Carr
- Mar 2
- 7 min read
Updated: Apr 25
March 22, 2020. Vancouver, British Columbia.
The SkyTrain should have been packed. At seven in the morning on a weekday, downtown-bound cars are normally shoulder-to-shoulder; commuters are braced against each other, headphones in, eyes down. Instead, I was nearly alone.
The silence was not pleasant at all.
It was the kind that makes you aware of every breath behind a mask. COVID had arrived. Not just speculation, but fact. Essential workers were still needed, and I was one of them; an IT consultant who enjoyed the work, loved the clients, loved the daily puzzle of printers and email servers and panicked calls about Outlook refusing to sync. That job was my way of learning Canada from the inside out.
But something else had already pulled at me.
For months I had been walking around the Downtown Eastside after work, introducing myself to overdose prevention tents, trying to find a way in. I had been in long-term recovery from a severe substance use disorder for years, and I had come to believe, the way many people in recovery do, that part of sustaining your own survival is helping others, whilestill standing at the edge. I was also one semester into seminary. I had no institutional footing yet. No established ministry. No clear path. So I looked elsewhere.
That is how I met Diana. Every afternoon she sat outside the Tim Hortons in her wheelchair. Funny, sharp, suspicious, and completely woven into that block. People greeted her, checked in with her. She knew who was struggling, who had disappeared, who was drifting. Her attention did not come from training or theory. It came from repetition, love, and proximity. One afternoon she stopped me cold. "Who are you? I see you around here all the time, but you're not getting high."
She had already made up her mind that I was either a cop, a religious nuisance, or both. I offered to explain over coffee and doughnuts. She agreed. I told her my story: recovery, IT, seminary, the strange collision of all those worlds inside one person. She listened the way only certain older women listen, with one eye on you and the other on the whole street. Then she said something that rearranged me.
"You can't talk about Jesus with the folks down here. They've heard it already."
She was right. Faith language did not automatically provide comfort in these spaces. For many people, Jesus was a trauma. The Church was a trauma. The 12-step slogans that once sounded like survival could land as intrusion, or judgment. I asked what I should do instead. She waved toward a young man across the street holding a glass pipe, slowly rolling it against a lighter flame. As he exhaled she said, plainly and with compassion, "You need to help our people like that."
Our people.
Diana had never struggled with substance use herself, but the consequences surrounded her and she refused to let anyone in her neighborhood become invisible. She did not speak about "those people." She spoke about our people. That phrase broke something open in me. It was also, though I did not yet know it, the design principle that would shape every technical decision I made in the months ahead.
A few weeks later, COVID shut down my consultancy. The job I loved vanished. But the loss created unexpected space, and I committed to a year of service in the SROs, the single-room occupancy hotels lining the Downtown Eastside.
I was not prepared.
From the outside, SROs read as bare-bones but manageable, old hotels repurposed into last-chance housing. Fifty dollars a month. A sink. A lock. Shared facilities.
The photos online showed cleaned-up exteriors. The reality inside was fluorescent bulbs flickering in stairwells, hallways full of sounds I had never heard collected in one place: moaning, sudden impacts, doors slamming, and people openly using substances in ways that, even given my own history, still caught me short.
What surprised me was not temptation. It was recognition. A clear awareness that without the support systems I had access to, treatment, recovery rooms, people willing to answer the phone, I could have been in one of those rooms myself. No one believed me when I said that. They could not picture the person standing in front of them ever having been in the grip of substances. I looked too put together. Too educated. It took old photos and videos, me skeletal, hollow-eyed, slurring into a camera, to make me legible.
Those images were not heroic. They were proof. In a place where people are constantly lied to by outsiders, proof was the only currency that bought trust.
Naloxone training came on my first shift. At that time we still used intramuscular injections rather than nasal kits. I had always been afraid of needles, which was darkly humorous given the situation. Two people needed Naloxone that first day. The first was unresponsive, body blue, not breathing. I found myself panicking, and worse, hesitating. A voice inside cut through: "it's your discomfort or their life." I steadied myself, remembered my training, and acted. I plunged the needle into a thigh and a few seconds later they drew in a ragged breath.
There is nothing quite like realizing your own discomfort stopped mattering about thirty seconds ago. In the weeks that followed I kept running into the same thing that trips up a lot of well-meaning people in frontline work.
When someone's life has been fractured long enough, the mess isn't a character flaw. It's what trauma looks like from the inside, what scarcity looks like, what surviving looks like when you've had to build something out of nothing and keep rebuilding it. We did room checks all day. Sometimes quiet is the problem. And Diana kept coming back to me.
That phrase. Our people.
I wasn't sitting with it as something nice someone said. I was sitting with it as a question I couldn't answer yet: how do you actually protect someone without watching them? How can you help without becoming intrusive? Eventually that stopped being a philosophical question and started being a technical one.
A few months before COVID I had written a grant proposal for edge AI equipment. The original idea was pretty modest, a system to count churchgoers. When the equipment showed up I had a different problem to solve, so I repurposed it.
I built an NVIDIA powered AI edge device using a MIT open source project as the base, added a FLIR infrared sensor and wrote some custom code on top of it. The job it needed to do was simple: check temperatures and confirm mask use before anyone came through the door.
Not a grand vision. Not a polished startup pitch. Just a need. The setup was as straightforward as I could make it. A hidden access point running a small local server, a second device in the staff office acting as the interface.
Someone walks up to the entrance, the sensor reads their temperature, the AI checks for a mask. Both clear, green light on the office screen, staff buzzes them in.
Ugly. Gloriously ugly. Built by someone under pressure who needed it to work, not to look good.
The hardware wasn't really the point. The privacy architecture was.
No facial recognition. No stored images. No data sitting somewhere on a server about people who never signed up to be in a database. It checked, it signaled, done. People came and went exactly as they always had.
That wasn't an accident. Communities like that one have had a lot of people show up with clipboards and good intentions and walk away with data. In harm reduction you figure out pretty fast that nobody owes you their trust. I talked through the privacy design directly with residents before anything went live. That wasn't a formality. Some of the hardest questions I got about how the system worked, what it kept, who could access it, came from those conversations. Not from any technical advisory process.
The box itself was solid. When it went down, you unplugged it and plugged it back in. That was the maintenance plan. Not a glamorous piece of kit. It didn't need to be.
By the time we took the system down in 2022, our building had one of the lowest COVID infection rates among SROs in the area. A small number of hospitalizations. In that context, that was remarkable.
I had done far more impressive technical work in Hollywood. More expensive systems, more elegant interfaces, more complex deployments. But this was the first project where my old technology life and my new life in service truly fused. Not "technology for good" as a conference slogan, but as something real. No product launches. No roadmap. Just a human need and a tool built to meet it.
The lesson that stayed: the most consequential technical decisions I made were not about sensors or code. They were about what the system was unwilling to do. It refused to store. It declined to identify. It refused to treat vulnerable people as data points. Those refusals, those deliberate absences, were the architecture that earned trust. And trust was the only reason the system worked at all.
After COVID passed, I never saw Diana again. I did not have her number. In a neighborhood like that, people vanish from your daily life suddenly as they enter it, carried elsewhere by illness, housing, family, crisis, or grace.
The rumor was that she had relocated to live with her family. I hope that is true.
I built something that kept people safe. I am grateful for that. But the deeper work of that season was done by people like Diana. People whose names will never make headlines, whose ministries will never be professionalized, whose theology was lived before it was ever spoken.
Sometimes the saints sit in wheelchairs outside coffee shops and call the neighborhood to order. Sometimes they hide their holiness in ordinary conversation, street-level compassion, and the simple refusal to let others become invisible.
Comments