Gettin’ Sparky with It: Building an Integrated Machine Learning Pipeline

This talk was held on Wednesday, March 13, 2019.

Data scientist/engineer duo Bryan Davis and Avadh Patel present systems that they developed at Indeed to aid in the training and deployment of machine learning models. Attendees left with a better understanding of the challenges data scientists face bringing their models to a production setting. Bryan and Avadh explore possible solutions that arise from data science and engineering collaboration.

Audio Description

The following video includes a descriptive audio track for this talk.

Transcripts

• Basic transcript (includes audio information only)
• Descriptive transcript (includes audio and visual information)

Speakers

Bryan Davis is a data scientist on the Jobs Models team. He joined Indeed in 2015 after completing his MS in Statistics and Operations Research in North Carolina. At Indeed, he designs optimal bidding strategies for marketing campaigns and builds models to predict the performance of jobs on Indeed’s website. He has also been a pioneer at Indeed in the use of Spark for data munging and model building.

Avadh Patel is a software engineer on the Jobs Models team, where he focuses on building products that integrate machine learning models. He joined Indeed in late 2018. Before that, he was at Oracle Research Labs identifying performance bottlenecks in databases and deep neural-nets for next generation CPU designs. He holds a Ph.D. in Computer Architecture. At Indeed, Avadh optimizes machine learning model deployment process to increase the rate of new model deployments and dynamically select a model based on input features.

Originally published at Indeed Engineering Blog.