COSS Community

loading...
Cover image for OCS 2020 Breakout: Jeff Pasternack and Huiji Gao
COSS Community

OCS 2020 Breakout: Jeff Pasternack and Huiji Gao

jj profile image Joseph (JJ) Jacks ・2 min read

Jeff Pasternack is a NLP Research Scientist at LinkedIn. Before joining LinkedIn, Jeff worked in a similar role at Facebook. Jeff received his PhD in Computer Science from University of Illinois.

Relevant Links
LinkedIn

Huiji Gao is a Senior Engineering Manager at LinkedIn leading the AI Algorithms Foundation team, responsible for developing horizontal AI technologies for search and recommender systems w.r.t. Natural Language Processing, Ranking, and Personalization. Before joining LinkedIn, Huiji was a research intern at IBM Research Almaden, and an engineering intern at LinkedIn. Huiji received his Ph.D. in Computer Science from Arizona State University under the supervision of Dr. Huan Liu, with research focus on data mining and machine learning.

Relevant Links
LinkedIn

Linkedin OCS 2020 Fireside Chat: Jeff Pasternack, Research Scientist and Huiji Gao, Senior Engineering Manager - Machine Learning and AI

A quick question - can you both provide background intros on yourself? What are you really working on at LinkedIn? - 0:00

I have to say, Jeff, the smart reply has probably saved me years of my life on LinkedIn. Thank you for that, it’s awesome! - 2:22

I want to kick off with a broad question that touches on the work you’re both doing at LinkedIn, which is digging into how NLP and AI apply to conversational data. How does LinkedIn solve these large scale NLP challenges? - 2:50

Have you noticed other internet companies using DeText and what types of use cases have you seen from the open source community? - 7:42

I’m curious what Jeff’s thoughts are on the evolution of NLP and how your current work is pushing this evolution? - 8:59

I’m curious about hardware abstractions at the silicon level catching up with some of these computation problems? Are we likely to see lots of custom chips built specifically for NLP? - 11:39

Maybe segueing on some of the challenges. What are the biggest tricky areas and things that you are focusing on, driving a lot of innovation in NLP? - 13:16

Maybe a bit of a technical question - what are some of the most interesting algorithms in ML that are really valuable for inferencing and understanding this unstructured natural language data? (JJ talks about playing with LinkedIn auto-reply suggestions in chat, trying to have entire conversations just by auto-replying) - 16:44

(Closing remarks) - 21:54


Share your questions and comments below!

Discussion (0)

pic
Editor guide