I got extremely drunk one day and absolutely embarrassed myself in front of a bunch of people, then spent$100 on an Uber home, and left my phone in the cab.
When I got home, I was so pissed off that I lost my phone I went into a blind rage, and in doing so I left the door open to my bus and my cat got out. When I woke up several hours later, she was gone.
I have a problem with drinking. I don’t know if I have a drinking problem, but I definitely have a problem with drinking. I don’t need alcohol, and I don’t even always want alcohol, but when I drink for one reason or another I cannot stop. That night it was probably 20 beers and 20 shots. A good night if volume is the measure of “good.”
The day was a mess. I was supposed to release my “Split the G” app - an app I’ve been dubiously working on for weeks. Essentially you take a swig of your Guinness, and if it lines up with the crossbar of the letter G on the pint glass, you get a high score. I spent all my time trying to get the calculation right. I started using Roboflow with an existing vision model, but as time went on anything less than perfect lighting produced inconsistent scores. It was easy to game. It wasn’t right. So I tried to train the model, but that takes AGES and possibly hundreds of beers, which normally wouldn’t be a complaint but I wasted so much time trying to build the scoring part I ran out of time to do that.
So I tried to introduce AI into the model. Claude Opus 4.6 has a fantastic vision model, but it kept getting confused in the backend, and it was extremely difficult to pinpoint why it was happening. Essentially the model thought the FOAM line was where you score, but the score is actually at the meeting of the foam and the beer. Since the model had no memory from subsequent runs, it sometimes would do it correctly, and other times it wouldn’t. No matter what I did linguistically to tell it that it simply couldn’t grasp it. So I went with a hybrid approach of using Roboflow as main with Claude Opus to be a fallback, and applied a heavy saturation to the image (which already had to be transformed to black and white for the vision model). This was decent, but the response time took FOREVER. So you could be waiting a minute for a score. Not the best experience.
Anyways, through all of that and with a time crunch I then had to build a live competition model for it. Without much direction I decided we would have everyone sign up in the app and we’d have a live updating leaderboard that showed signups as they came in, which was cool, but it was 1 step forward and 10 steps back. I then spent so much time with database problems with Supabase (which I had never used before) that it ate all of my time, and I never got the competition flow working.
You can guess how it went.
The competition launched and the responses took over 30 seconds so some people timed out and others failed and refreshed. It was over as soon as it began. People left angry and there was no round 2 for me. It then went to a manual mode, and I was so embarrassed that I decided FUCK IT and started heavily drinking. Then I fell asleep on the couch at the bar after professing like for someone, which I don’t even know why I did that.
So that was that. And then my cat got out and everything was my fault. Hubris, maybe. Recklessness and a complete lack of situational awareness, as well as the inability to bring the project to fruition all played their part.
The app still exists, but I still want to tweak it more. The people who paid for the app haven’t asked about it. They don’t care, because it failed. There won’t be a 2nd attempt to run a competition like that.
Since then, I have done everything. Walks at night, walks during the day, walks whenever. Looking for her, leaving her things outside, leaving my things outside, nothing. She hasn’t been found. It rained a few nights which didn’t help. I came back daily to set up food, because through all of this I had to move out of my bus and into a hotel somewhere else. So I’ve been taking 30 minute drives back to where she last was hoping to find her, but nothing. I posted on all the apps, I even called a “pet detective” who never called me back. I put up flyers. I phoned every vet. I’m almost out of ideas….except…
I have a DJI Mini 3 Pro and I tried scouring the neighborhood for her, but that was a farce. During the day in Miami any self respecting cat would be inside, or in shade and hidden. So I thought to myself…what can I do here? I grabbed 5 minutes of video and looked all over, and decided I would try to build a vision model to scan for potential cat hiding places, food sources, water sources, or anomalies.
So far it’s a theory that it could help, but with the DJI data I could then draw a map with pins of where to look and give it a try.
But really, I just think someone took her in and decided not to look for the owner. So behind that, I’m creating mailers, gathering the names of everyone who is a home owner in the area, and I’m going to mail them a letter asking if they’ve seen my cat. I’ll set up a hidden phone number for them to call and text, and an anonymous email for them to give tips to. I think that might help.
But who know? Not me. I miss her so much.