Now, 13 million swipes later, it finally makes sense to train a custom built recommender
Shipping fast isn’t always about coding fast, but sometimes building less
First time CTOs don’t realize that their lack of experience is their biggest strength
At @styl_app’s inception, rather than spending months learning to build the perfect fashion recommender, I spent all my time shipping incrementally smarter MVPs with just an off the shelf model
If someone sees the @styl_app repo they might actually cry…
It was my first react native project and some of the files are like 1900 lines. Spending the weekend migrating to tsx and cleaning it all up, debugging takes way too long 🤦♂️
If someone sees the @styl_app repo they might actually cry…
It was my first react native project and some of the files are like 1900 lines. Spending the weekend migrating to tsx and cleaning it all up, debugging takes way too long 🤦♂️
In 2022, H&M offered $50k to anyone who could build a better fashion recommender.
Over 3700 PhDs, research scientists and professors worked relentlessly to win.
@10rbhandari and I are students @DukeU. This is the story of how we beat them all to create @styl_app🧵👇 https://t.co/3dMmE7kYxu