Let the listener choose.
When the system isn't sure which track or version fits, show candidates clearly.
DEAR SPOTIFY TEAM,ONE PRODUCT PERSON · ONE REAL PROJECT
THIS ONE'S PERSONAL
I want to
build at Spotify.
So I built something to show you how I think.
I'm a product leader with experience at consumer scale and in enterprise AI. I'm also a music obsessive. When I noticed a gap between discovering a song and keeping it, I started building Needle.
Watch the 39-second demo REAL PRODUCT / REAL DECISIONS
MY IPHONE / ACTUAL BUILDWATCH THE 39-SECOND DEMO ↑
Independent personal work. Neither Skaromny nor Needle is affiliated with or endorsed by Spotify.
01 / THE REASON
The best discoveries
don't always happen
inside a music app.
Sometimes it's a Reel, a YouTube mix or a clip from a friend. I kept finding music I loved and losing it on the way to my library. Needle started with that very ordinary frustration.
02 / THE PRODUCT JUDGEMENT
The important part
wasn't recognition.
Anyone can demo an AI result. I cared about what happens when the result is uncertain, slow or almost right.
Use intelligence deliberately.
Start with cheap, reliable signals. Escalate to expensive inference when needed.
Keep the flow human.
The aim is a playlist worth saving, with control over the final selections.
REALITY CHECK
In the filmed example, 22 tracks were selected. It's a demonstration of a working flow, not a production performance claim. Waiting periods were edited out.
03 / WHY I'D BE USEFUL
A builder's instinct.
Experience at scale.
Needle shows initiative. My background brings the other half: experience making decisions with real users, teams and business outcomes.
16MMAU OF THE MEGAFON APP I LED
+1MMAU GROWTH DURING MY LEADERSHIP
AIPRODUCT WORK AT VODAFONE GROUP
I'd love to make
the next thing together.
I'm interested in the problems Spotify teams care about: discovery, engagement, trust and the everyday listening experience. I'd love the chance to contribute.
Let's talk