Allen Tran

I'm a Principal Machine Learning Scientist at Netflix. My work sits at the intersection of reinforcement learning, causal inference, and behavioral/preference modeling — building rigorous, bottom-up models of how people actually make decisions, and shipping them at scale. I care about problems where the science is genuinely hard and the stakes are real.

Selected research

• the value of personalized recommendations: evidence from Netflix (EC 2026)
• inferring the long-term causal effects of long-term treatments from short-term experiments (ACIC 2024, ICML 2024 oral, ~top 1% of submissions)
• beyond customer lifetime valuation: measuring the value of acquisition and retention for subscription services (WebConf 2022)
• deep learning for driving detection on mobile phones (KDD 2016)
• causal inference on observational studies @ Hulu

Projects

• fed translator - predict paths of the fed funds rate from Fed statements/minutes
• graph2vec - learn vectors for nodes in graphs
• use all the data in forecasting - probabilistic PCA with missing data
• traffic speeds by zipcode
• uncovering the structure of the US economy via production/consumption networks
• propagating recommendations via the social graph

Slides on various papers (other authors):

• Memory Networks (Facebook AI Research)
• Deep Neural Decision Forests
• AlphaGo (DeepMind)

Fun proofs (I wish they were my proofs, they are just my notes!)

• Dinitz / Graph Coloring Proof from "The Book"
• 6 proofs of the Infinitude of Primes from "The Book"

My earlier econometrics research from my Econ PhD at UCLA is here.