Shaped

Shaped allows you to control exploration and diversity with ease ??

We're hiring!

Shaped connects directly to customer data platforms like Amplitude, so you can directly ingest your data without having to set up tedious logging infrastructure. https://lnkd.in/gJkqdmYw

Setting up Shaped just got even easier ?? Now you can easily upload your datasets and models directly to the dash. Try it here: https://lnkd.in/eaundxsr

Good news! Shaped now supports ingesting data from Apache Iceberg tables. Read the docs here: https://lnkd.in/gAMpiYqj

Diversity factor defines how much to trade-off relevance and attribute diversity when re-ranking results. We achieve this with a Maximal Marginal Relevance algorithm that chooses the most diverse item out of a set of maximally relevant items at each rank. https://lnkd.in/gD-EurqV

??Have you got data drift? ?? ?? It might be due to more than just your machine learning models. But what exactly is data drift? Data drift is like a GPS that gradually becomes outdated. Imagine you're using a GPS to navigate a city, but the city is constantly changing—new roads are built, old ones are closed, and traffic patterns evolve. If your GPS doesn't update, it will give you wrong directions because it's relying on outdated information. Similarly, data drift happens when the data used to train a machine learning model changes over time. If the model isn't updated to reflect these changes, its predictions become less accurate, much like the GPS leading you astray When probability of sale scores started dropping at Stitchfix, the team scurried to dissect their models for answers, they needed to find what was broken fast. But after digging deeper, they realized the mismatch came from their marketing efforts, not their models. See our entire interview on building recommendation systems at scale with Jason Liu here: https://lnkd.in/gPDpA3TV #DataDrift #MachineLearning #AIinbusiness

You can now directly connect your S3 buckets to a Shaped dataset, allowing you to flexibly ingest data in a variety of formats automatically. https://lnkd.in/gV_xGCqC?

?? Video feeds like Instagram Reels and TikTok are made or broken by their ability to deliver highly personalized content in real time. For video commerce apps like sune, a livestream shopping app by QVC, personalization is just as, if not more important, as users already have an expectation of the types of content and products they'd like to see. Instead of hiring an entire team to build a recommendation system from scratch, sune partnered with Shaped to bring a highly personalized shopping experience to their 'For You' feed in less than one sprint. Read our latest case study on Sune here: https://lnkd.in/gG-QDtTx

How do you optimize a recommendation system for happiness? ?? For most companies, prioritizing shifting the needle for user engagement is a no-brainer, but what about for a meditation app like Headspace, where the end goal is something more intangible like improving mindfulness? We recently chatted to Rohan Singh Rajput, Personalization Lead at Headspace, to learn more about how their team has built a system with the aim to make the world a happier place.