Wednesday, October 20, 2021 - 1:00pm to 2:00pm
Zoom: https://oregonstate.zoom.us/j/93591935144?pwd=YjZaSjBYS0NmNUtjQzBEdzhPeDZ5UT09

Speaker Information

Javad Azimi
Machine Learning Engineer
Twitter

Abstract

Few years ago, I was sitting at the other side of the table at OSU where I was eagerly and happily listening to the people from outside of OSU sharing their research with amazing results. Top notch researchers who moved (and are moving) the edge on science; I am not one of them now. I always wanted to go through the same path but I finally convinced myself to change the gear and move toward the product side in tech companies rather than research. World is much different here as the challenges are moved from ML sides to the product sides. How a problem is being defined and brought into our attention? How do we measure and evaluate the potential impact of allocating resources (human and budget)? How to prototype and make conclusions based on it?  How to handle the scale and pave the path into having a product? And how to maintain and keep it up and running? In this talk, I am going to share my experiences in all of these steps with you with an interesting example: Misinformation Detection. My goal is to give you a taste of work waiting for you if you decide to go through the same path as I did.

Speaker Bio

Javad Azimi is currently a tech lead ML engineer at Twitter working at the Misinformation team where the goal is to reduce the prevalence of misleading information at Twitter! Prior to joining Twitter in March 2020, he was serving the same role at Facebook with special focus on Facebook/Instagram posts containing photos and videos. Before Facebook he worked at Microsoft where he developed NLP models for ads retrieval team for 4 years. He graduated from the EECS department of Oregon State University under the supervision of Xiaoli Fern and Alan Fern in 2012. His PhD thesis title was “Constrained Bayesian Optimization” where they published a few papers at MLJ, NIPS, ICML, KDD, AAAI, IJCAI, ECML etc.