How Realtor.com implemented a production-level feature drift pipeline with Evidently AI 📚

📈 The purpose of the feature drift detection
📊 An overview of the Evidently tests they use
💻 A script that generates a full drift report

techblog.realtor.com/how-realtor-co…

10th annual MAD (Machine Learning, AI & Data) Landscape is out 🚀

We are thrilled to be featured in the AI Observability category 📷🔥📷

Thanks to Matt Turck уt al. for putting it together and mentioning Evidently AI!
mattturck.com/mad2024/

📌 In case you missed it

A tutorial on creating ML Model Cards in Python 🗂

Follow the code example to create and customize ML model cards using open-source tools.

You can adapt it to your use case and data 👇
evidentlyai.com/blog/ml-model-…

A Friday ML use case 📕

📚 From the database of 300+ ML systems: cutt.ly/SwrZWL0g

How Etsy uses real-time user actions to show relevant ads: diversifiable personalization module, pretrained representations & ADPM-personalized sponsored search 👇
etsy.com/codeascraft/le…

A Thursday ML observability lesson 👩‍🎓

💻 How to detect data and prediction drift: code practice

Follow this video lesson as we walk you through the code example of detecting data drift using the open-source Evidently AI Python library:
youtube.com/watch?v=oO1K4C…

Innovative recommendation applications ❤️

Uber shares how they use Two-Tower Embeddings (TTE) to enhance their recommendation system, the TTE model architecture and training strategy:
uber.com/en-GB/blog/inn…

📌 In case you missed it

Learn how DeepL implements their MLOps practices with:

🏗 Building MLOps platform with open-source tools
📈 Production ML monitoring: from EDA to data integrity checks to monitoring data drift

Read the full story:
evidentlyai.com/blog/how-deepl…

A Friday ML use case 📕

📚 From the database of 300+ ML systems: cutt.ly/SwrZWL0g

How Siemens Healthineers optimizes software testing: predicting test results at commit time and optimizing test suites 👇
infoq.com/articles/machi…

A Thursday ML observability lesson 👩‍🎓

Deep dive into data drift detection:
📊 Drift detection methods and how to choose the optimal approach
📚 How to interpret data and prediction drift
✅ What to do if drift is detected

Watch the video:
youtube.com/watch?v=N47SHS…

Scaling JupyterHub ecosystem to organization-wide deployments 🚀

Yelp shares the architecture of its JupyterHub ecosystem and how it evolved to become an easy-to-use, scalable, robust, and monitored system for all engineers on the team 👇
engineeringblog.yelp.com/2023/07/overvi…

📌 In case you missed it

A tutorial on monitoring unstructured data for LLM and NLP 🔡

Follow the code example to track interpretable text descriptors that help assign specific properties to every text.

You can adapt it to your use case and data evidentlyai.com/blog/unstructu…

A Friday ML use case 📕

📚 From the database of 300+ ML systems: cutt.ly/SwrZWL0g

How Salesforce summarizes Slack conversations: they built AI Summarist, a conversational AI tool that summarizes conversations and identifies important threads 👇
blog.salesforceairesearch.com/ai-summarist-s…

A Thursday ML observability lesson 👩‍🎓

Data & prediction drift in ML:
📊 Key concepts of data & prediction drift
⚠️ Drift detection methods: statistical tests, distance-based metrics, rule-based checks
✅ How to choose the right method

Watch the video:
youtube.com/watch?v=bMYcB_…

🏗 Maintaining own ML stack for each ML team?

It takes immense engineering resources and can be a major bottleneck for ML innovation.

Pinterest shares how they fixed the problem by standardizing ML under one ML engine 👇
medium.com/pinterest-engi…

📌 In case you missed it

🔢 Is your data drifting?

Here is a deep dive into data drift and how to detect it:
📊 Data drift vs. concept drift vs. prediction drift
⚠️ How to detect data drift
✅ How to handle data drift

evidentlyai.com/ml-in-producti…

A Friday ML use case 📕

📚 From the database of 300+ ML systems: cutt.ly/SwrZWL0g

How Whatnot detects marketplace spam: they use LLMs to enhance multimodal content moderation, fulfillment, and general fraud protection 👇
medium.com/whatnot-engine…

A Thursday ML observability lesson 👩‍🎓

💻 How to evaluate data quality: code practice

Follow this video lesson as we walk you through the code example of data quality evaluation using Evidently AI Reports and Test Suites:
youtube.com/watch?v=_HKGrW…

Mean Reciprocal Rank (MRR) explained ⚡️

MRR at K helps assess the ranking quality by considering the position of the first relevant item in the list.

This detailed guide explains how to compute and interpret the metric:
evidentlyai.com/ranking-metric…

Effective embedding applications

Nextdoor shares how they iterated on the development of embedding models:
🧭 From pre-trained to fine-tuned
✅ Featurizing and serving at scale into product applications
🧠 Challenges encountered during the process

engblog.nextdoor.com/from-pre-train…

📌 In case you missed it

How to detect drift in ML embeddings?

We compared 5 drift detection methods and tested how well each of them catches shifts in data.

Key findings and the code to reproduce the research 👇 evidentlyai.com/blog/embedding…