Instagram in focus

on YouTube ↗18m

" summary

Instagram handles about 80 million uploads per day and 400 million users while combining machine learning and human curation to improve discovery. Early heuristics evolve into a learned ranking with about 90 features running on Facebook ML platform, yielding concrete gains like 5x engagement from showing photos liked by people you follow and 7x improvement from showing photos liked by people whose photos you like. User bios power scalable K‑means clustering into editable categories, and live event stories pipeline filters millions to thousands for editors who tune ranking by real‑time engagement signals.

AI-written summary · about

" chapters (6)

1. 0:00Introduction and history
2. 2:42Early editorial curation
3. 5:16The popular page
4. 7:15Evolving algorithms
5. 10:06Categorized user discovery
6. 13:52Live storytelling

" related talks

• 48mFrom Filters to Stories, and the Evolution of InstagramRodrigo Schmidt · QCon São Paulo 2019
• 46mFacebook & Instagram: Strategies, Perspectives and Trends 2022Simon Beyer · Digital Marketing Europe 2022
• 22mGenerating a Billion Personal News FeedsEwa Dominowska · MLconf 2016 Seattle
• 19mJourney to 1000 Models: Scaling Instagram's algorithm without the Reliability NightmareSing Sing Ma, Luke Levis · @Scale: Systems & Reliability 2025
• 09mSocial Media Image Analysis for Public HealthVenkata Rama Kiran Garimella, Abdulrahman Alfayad, Ingmar Weber · CHI 2016
• 42mModeling Aesthetic Identity- Building a Digital TwinPranav Kompally · PyData Boston 2025
• 03mServing Billions of Personalized News Feeds with AIMeihong Wang · Artificial Intelligence Conference 2018 New York
• 25mWhen design meets machine learningLalo Martínez · Config 2021