The Fifth Elephant 2017

On data engineering and application of ML in diverse domains

From a recommendations carousel to personalizing entire app - personalization story at paytm

Submitted by Charumitra Pujari (@charupujari) on Tuesday, 4 April 2017

videocam_off

Technical level

Advanced

Section

Full talk in Payment Analytics track

Status

Confirmed & Scheduled

View proposal in schedule

Vote on this proposal

Login to vote

Total votes:  +10

Abstract

At paytm we value user experience and we want to pre-emptively show a user the types of products they would want to buy. In this talk, we will walk our audience through how we personalize every pixel on our app. How do we use deep learning on tens of terabytes of data everyday to sort long tail merchandise and how we use an ensemble of several models to generate every recommendation. We will share our learnings from trying several iterations of models and we will show why standard recommendation techniques are not widely applicable and why we had to come up with our own framework for solving these problems.

Outline

At paytm we value user experience and we want to pre-emptively show a user the types of products they would want to buy. In this talk, we will walk our audience through how we personalize every pixel on our app. How do we use deep learning on tens of terabytes of data everyday to sort long tail merchandise and how we use an ensemble of several models to generate every recommendation. We will share our learnings from trying several iterations of models and we will show why standard recommendation techniques are not widely applicable and why we had to come up with our own framework for solving these problems.

Speaker bio

Charu is a seasoned machine learning and big data technology leader with over 11 years of experience
building data products for companies like Amazon, Canadian Tire and Accenture.
At Paytm, Charu manages data product teams like personalization, seller scoring, customer scoring and
product forecasting.
Charumitra is a hands on leader with experience leveraging machine learning in many verticals like
personalization, retail merchandise planning, loyalty marketing, digital analytics, supply chain modeling
and operations optimization. He has expertise in setting up data science teams and scaling them to
automate decision making using machine learning.
Prior to Paytm, Charu, setup first data science team at Amazon in Canada. Prior to that he was Solutions
Architect at Canada’s largest retailer helping them setup big data and analytics team responsible for
managing assortment across 1000+ stores.
Charumitra Pujari has graduated from the McMaster University with a Masters in Engineering.

Slides

https://www.slideshare.net/secret/4hE1Icj9u9GV21

Comments

  • 1
    Zainab Bawa (@zainabbawa) Reviewer a year ago

    Is this framework open source? If not, do you intend to open source it?

Login with Twitter or Google to leave a comment