How a Headbanging Engineer Became a Vibe Composer

When I was a kid, I would make up tunes on my saxophone with a whole band playing in my head.

Nothing was written down. I don’t remember any of those tunes. I remember enjoying myself.

I played in concert band at school and in jazz band, and I practiced a lot on my own for fun. I loved freeform jazz. I still love hearing a good saxophone in a song or a live band. But when I was making up those tunes, I wasn’t thinking about composing, recording, or finding other musicians. I was too young to worry about any of that. The band was already there in my imagination, and I was just enjoying the tune.

Music has always had that kind of hold on me. Even as a young boy, I loved rock and metal. I’d bounce around the living room, swinging around like I was in a mosh pit. When I found a great song, I could listen to it over and over. Then I’d go looking for another artist, another genre, another sound worth getting obsessed with.

Repeating a song and exploring new music were never opposites to me. Both came from wanting more of that feeling.

Years later, I’ve found a new way to get it.

But the route there went through software engineering.

Moving up a layer

As an engineer, I’ve always worked with things I didn’t build.

I used libraries maintained by other people. I trusted the tools that translated my code into instructions a machine could execute. I trusted hardware I hadn’t designed to execute them correctly.

Even a simple loop I wrote in Python to count items in a list depended on an enormous amount of work beneath it. I could write a few lines and get a result because other people had already made it possible for me to work at that level.

I didn’t consider the program any less mine because I hadn’t personally built everything it depended on.

Later, moving from founding engineer into engineering leadership, I had to make a related adjustment. I couldn’t personally implement everything. I had to discuss what we were building with my engineers, trust them to produce it, and judge whether the result accomplished what we intended.

That made the shift to AI-assisted coding feel less alien to me than it might have otherwise. I was already comfortable separating responsibility for an outcome from personally executing every step.

There were still decisions to make. There was still a difference between something that ran and something that was right. But I could move more of my attention toward what I wanted to create, rather than the mechanics of producing every part.

Then I found myself doing something similar with music.

The first song sucked

I heard about Suno through AI influencers. My love of music drew me to it, but I spent a couple of months thinking about trying it before I actually made anything.

I knew enough from working with AI to expect rough edges. Getting a result is one thing. Getting a result you’re happy with usually takes more work.

Eventually, I used ChatGPT to help develop a story, lyrics, and a Suno prompt. I put them into Suno and left the settings at their defaults.

It was a metal song about stoicism.

Compared with the music I make now, that first song absolutely sucked.

Listening to it was incredible.

Those two things are both true. I could hear its shortcomings, but I was also hearing my story come back as a complete song. Something I had wanted to express now existed in a form I could put on and listen to.

The power of that instrument was astonishing.

I started changing things. A lyric, a prompt, another attempt. That first song wasn’t the finished destination. It was enough to make me keep going.

For the first Amor Mori album, I worked through 252 Suno outputs to arrive at seven tracks. By my estimate, that’s about 50 hours of listening alone.

The full album is ready. I’m releasing it one song at a time, starting with “Lie With Dogs” on October 9, then a new track each month.

Now, when I make music for Amor Mori, the first listen is still where the rubber hits the road. Sometimes a song flops. I hear it and think, No, this just didn’t work. Other times it rocks, and I’m immediately interested in making the small changes that might get it the rest of the way there.

But the moment that matters most is the same one I’ve been chasing as a listener for years: hearing a song for the first time and falling in love with it.

The unusual part is that I can have that experience with a song I helped create.

Vibe composing

I’m often describing a feeling or a story rather than specifying every musical detail.

Sometimes I want to hear something that resonates with an emotion I’m feeling right now. Sometimes I’m imagining a character and what they’re going through. I want the music to inhabit that experience.

I bring the intention, work on the lyrics and direction, and then get to hear what happens when those inputs become a whole song.

I’m not assembling every piece individually, so there is still a discovery waiting for me. I can know what I was trying to make without knowing exactly what I’m about to hear.

That combination is what makes it so exciting. I’m directing something, but I’m also its first listener.

Vibe coding had become vibe composing.

Suno isn’t simply a compiler for music. It makes musical choices I haven’t specified. I’m not claiming that describing a song is the same physical or technical act as playing a saxophone, or that I personally performed the instruments and vocals in the result.

The analogy, for me, is about moving up a level of abstraction. I’m comfortable working through a system that handles more of the implementation, while I concentrate on intention, direction, and whether I actually like what comes back.

I don’t experience that as a reduction in authorship. I experience it as an expansion of what I’m able to create.

An idle saxophone makes no music. An unused Python terminal produces no program. An idle Suno account makes no songs.

Someone has to use the instrument.

That is the beginning, though, not the whole job. Having an intention doesn’t guarantee a good result. Neither does typing a prompt. My first song is evidence enough of that.

Taste matters. So does taking action: making something, listening to it, deciding what works, and doing something about what doesn’t. There is a difference between having access to an instrument and using it to produce music you care about.

To me, that is where the authorship lives. Not in a count of how many implementation details I personally executed, but in bringing an intention to the instrument and exercising judgment over what I make with it.

I see that as democratizing creative ability. More of what I want to express is within reach. That doesn’t make every result worth listening to, and it doesn’t mean everyone else has to like my music.

A song still has to make me want to hear it again.

At least in my own head

The biggest surprise has been how often mine do.

I don’t just listen to them while I’m working on them. I put them on because I enjoy them. They go on repeat, the same way someone else’s song would when it caught me.

There’s something familiar in that pleasure. As a kid, I could enjoy making up a tune without needing to explain what I was doing or decide whether it counted as anything. Now I can make something that stays around, something I can return to as a listener.

And sometimes it comes back without my asking.

A lyric or a hook will get stuck in my head. Not the whole song, just that one part, going around and around. I’ll be trying to place it.

Shoot, who is that? What song is that?

Then it hits me.

Holy shit. That’s my song.

That is the most exciting part of all this. For a moment, I’ve forgotten where the music came from. I’m not thinking about prompts, software, or authorship. I’m just a person with a song stuck in his head, wanting to hear it again.

I’m alongside the greats, at least in my own head.