Why I Build This Way

A year ago I nearly died from alcohol dependency. Ambulance. Five days in the ICU. When I came out, I'd lost most of my relationships and my daily life.

I moved to North Carolina and spent eight months rebuilding from a bedroom — one measured day at a time.

I'm now over a year sober.

The constraint

I also have a perforated right eardrum — about 50% hearing loss on that side. During recovery, I decided to learn to sing. I was writing an album called Gravel & Grace about the whole arc, and I wanted to record it myself. But I couldn't reliably hear whether I was hitting the notes.

So I did what I do: I built a system.

vocal-tools is a 70+ metric Python pipeline that analyzes vocal recordings the way a producer would — pitch accuracy, tonal clarity, dynamic expression, formant placement, presence. It replaced the feedback loop my ear couldn't provide.

Over ten months and 323 analyzed takes, that pipeline taught me something about myself. My pitch problems weren't talent problems. They were monitoring problems.

When I switched to bone-conduction monitoring that reduces reliance on the damaged eardrum, my intonation improved overnight. When I stripped the backing track down to a single sine-wave drone, singing a cappella with that drone produced the same intonation as singing over the full track.

The constraint wasn't the limit. It was the catalyst.

The pattern

Once I saw the pattern, I couldn't unsee it. Three times now, a constraint has produced something genuinely new:

Three products. Three constraints. One pattern: the constraint creates a gap. You can accept the gap or close it.

Why proof

Everything on this site — the audit trails, the integrity checks, the test reports, the vocal analysis plots — comes from the same place. I don't take anything on faith anymore. I measure it, verify it, and ship something real.

That isn't just an engineering philosophy. It's how I got sober. Small loops. Measurement. Consistency. One day at a time, with evidence that the day actually happened and the work actually got done. The recovery and the engineering run on the same operating system.

The pipeline even taught me when to stop looking at it. After an hour of declining scores, I closed my eyes and stopped thinking about metrics. The next take jumped 10 points. The numbers train the body. The body performs without them.

I earned a 3.97 GPA in my M.S. in Computer Science while I was drinking and barely studying. I scored a 19 on the ACT — a standardized test I took with undiagnosed hearing loss and no accommodations, in a measurement environment that didn't match how my brain actually works. I believed the wrong score defined my intelligence for twenty years.

The data is how I trust what my emotions sometimes won't let me believe. When the numbers say I'm good, I can start to accept it — because the numbers don't lie and they don't flatter.

What it looks like now

I'm in Illinois. I'm working as an SRE at U.S. Bank. I'm recording Gravel & Grace. I'm learning guitar. I share the journey publicly on Snapchat — singing videos, recovery updates, honest moments of rebuilding — not for attention, but to show people what recovery actually looks like.

The singing is not separate from the recovery. It is the recovery. Every take, every session, every improvement is me rebuilding myself and proving — to myself and to the people watching — that forward motion is real and measurable.

I don't need to be told I'm great. I need to be told the truth. The data is how I trust it.

If you've read this far, you know more about me than most people do anymore. The ones who stuck around are few, and they matter deeply. If you want to build something together, reach out.