Ihza MahendraResume
Back to Production ML at scale
Production ML at scale

Recommendation engine for the TikTok Shop landing page

Recommendation engine that ranks which products a shopper sees first. +129% GMV.

What it is

A recommendation engine that decides which products to surface, and in what order, on the landing page of TikTok Shop.

What it's for

When a shopper lands on the page, the platform has milliseconds to decide which products greet them. Getting that decision right is the difference between a purchase and a bounce. This engine replaced the previous one on that surface and moved much more revenue through it.

How it was built

A deep-learning ranking model trained and served on ByteDance's internal cloud at production scale. The pipeline covered feature engineering, candidate generation, model training, and low-latency online serving, validated through online A/B testing against the prior baseline.

My role

Co-built as part of the Algorithm team at ByteDance.

Built with
Deep learningRecommendation systemsCandidate generation & rankingDistributed inferenceByteDance cloudPythonC++

Want the full technical depth, the tradeoffs, what broke, what I'd do differently? Ask the agent about this project.