Instagram ML Question - Design a Ranking Model (Full Mock Interview with Senior Meta ML Engineer)

In this ML System Design video, we ask a Senior Machine Learning Engineer from Meta to design a ranking and recommendation system for Instagram. He focuses on increasing user engagement by optimizing post suggestions from both friends and content creators, aiming to boost daily active users and session times by implementing AI in clever ways. Our guest explains the model's functional requirements, emphasizing predicting engagement actions like likes, comments, and views. He also covers the importance of MLOps tools for analytics, monitoring, and alerts to ensure the system's effectiveness and reliability.

Chapters (Powered by ChapterMe) -
00:00 - Designing Instagram's Ranking Model
03:24 - ML Model for Instagram Metrics
08:33 - ML Pipeline Nonfunctional Requirements
10:22 - Monetization Through Ads
12:03 - ML Pipeline Stages Overview
19:18 - Pretrained Embeddings for Interaction Analysis
24:04 - Comprehensive Model Pipeline Strategy
31:23 - Collaborative Filtering for Efficient Representation
33:13 - Two-Tower Network for Data Filtering
38:43 - ML Maturity & AUC Curve Analysis
44:58 - Microservices for Continuous Learning and Scaling

Want more machine learning content?

ABOUT US:
Did you enjoy this video? Want to land your dream career? Exponent is an online community, course, and coaching platform to help you ace your upcoming interview. Exponent has helped people land their dream careers at companies like Google, Microsoft, Amazon, and high-growth startups. Exponent is currently licensed by Stanford, Yale, UW, and others.

Our courses include interview lessons, questions, and complete answers with video walkthroughs. Access hours of real interview videos, where we analyze what went right or wrong, and our 1000+ community of expert coaches and industry professionals, to help you get your dream job and more!

[MeghavVerma]

The Meta engg guy is on-point. Every stage of the pipeline/process has so many nuances which could have made this into a 2hr+ video - maybe consider doing a podcast version of this and make a guestlist of viewers who can submit questions? @Exponent

[mulangonando]

This is perfect content for even guys who are just looking to activate mental faculty in fullstack ML design. The whole scale of thought process from concept to concrete algorithms is super transparent for many

[pranav]

The best ML design interview I have seen so far!

[abhipatel]

Very educational! Loved it, Keep on brining more ML interviews.. :)

[maryamaghili]

for the candidate generation, I would propose a funel model in which first I use some simple algorithms like logistic regression or Dtree or ANN which he used to quickly narrow down the search space to 1/1000 and then do more advanced techniques for refining it. I will use 2 tower network for ranking my candidates.

[soustitrejawad]

As a current master's student in data science actively job hunting, I must say this mock interview is incredible. Thank you so much, Vicram! Where can I find more of your content?

[_anarki_]

he's cracked. great job to you both

[RezaE]

This was a great mock interview. Thanks for sharing it.

[xgu]

One comment could be that two tower network should also be categorized as collaborative filtering

[EranM]

Where is the cycle of learning? How about monitoring? When do we train? Do we automate it? how?

[haoyuwang]

just out of curiosity, do you think the performance is good enough to pass a senior level MLSD interview?

[sharulathan]

can someone explain the label part in the two tower model ?

[PrudhviRaj]

Thank you so much for this video. I have a question. So once the two-tower model is trained, for candidate generation, the embeddings for items are computed offline, and the user embedding is computed on the go with the user features, and that user embedding is used against the item embedding vectors for kNN. Is that correct? If so, since the output of the two-tower model is binary, where would I be getting the embeddings from? From a layer before the sigmoid?

[MrAnujchopra]

As I understand, it is just 1 model for candidate selection and for ranking. Then why go to that same model twice? We generate post embedding asynchronously. Does Approximate Nearest Neighbour search is faster than taking dot product with all items.
Also 0.5 ROC-AUC is not random prediction but a constant prediction for all values.

[alexilaiho]

Great interview. I have always been confused, in an ML System design interview, should we focus on the ML model data/training/eval pipeline more or the inference pipeline( which is ore of a traditional system design) more ??