“Viral is very knowledgeable in the domain of programming and software design and has great attention to detail. One thing that stands out about him is the ability to visualize all possible future scenarios and design the system so that it is robust/fail-proof right from the start. Other than that, he's very helpful to his team members and also quite eager to learn relevant new methodologies. As such I would highly recommend him, he would be a great asset to any team that he is part of.”
About
Activity
-
🌟🌟🌟 Great initiative by Income-tax department, Indian Ministry of Finance 🥇🥇🥇 Most of tax payers in India feel that there is no recognition of…
🌟🌟🌟 Great initiative by Income-tax department, Indian Ministry of Finance 🥇🥇🥇 Most of tax payers in India feel that there is no recognition of…
Liked by Viral Gupta
-
Congratulations! Abhishek Mangal and EcoSheets Team
Congratulations! Abhishek Mangal and EcoSheets Team
Liked by Viral Gupta
-
My favorite measure of success for a company is the number of employees turned liquid millionaires. I realized we at Levels.fyi have some of the…
My favorite measure of success for a company is the number of employees turned liquid millionaires. I realized we at Levels.fyi have some of the…
Liked by Viral Gupta
Experience & Education
Licenses & Certifications
Publications
-
Rise of the Machines: Removing the Human-in-the-Loop
Operational Machine Learning
Most large-scale online recommender systems like notifications recommendation, newsfeed ranking, people recommendations, job recommendations, etc. often have multiple utilities or metrics that need to be simultaneously optimized. The machine learning models that are trained to optimize a single utility are combined together through parameters to generate the final ranking function. These combination parameters drive business metrics. Finding the right choice of the parameters is often done…
Most large-scale online recommender systems like notifications recommendation, newsfeed ranking, people recommendations, job recommendations, etc. often have multiple utilities or metrics that need to be simultaneously optimized. The machine learning models that are trained to optimize a single utility are combined together through parameters to generate the final ranking function. These combination parameters drive business metrics. Finding the right choice of the parameters is often done through online A/B experimentation, which can be incredibly complex and time-consuming, especially considering the non-linear effects of these parameters on the metrics of interest. In this talk we will present how we build generic solution to solve the problem at scale.
Other authorsSee publication -
Bayesian Optimization for Balancing Metrics in Recommender Systems
IJCAI (Rescheduled to 2021)
Most large-scale online recommender systems like newsfeed ranking, people recommendations, job recommendations, etc. often have multiple utilities or metrics that need to be simultaneously optimized. The machine learning models that are trained to optimize a single utility are combined together through parameters to generate the final ranking function. These combination parameters drive business metrics. Finding the right choice of the parameters is often done through online A/B…
Most large-scale online recommender systems like newsfeed ranking, people recommendations, job recommendations, etc. often have multiple utilities or metrics that need to be simultaneously optimized. The machine learning models that are trained to optimize a single utility are combined together through parameters to generate the final ranking function. These combination parameters drive business metrics. Finding the right choice of the parameters is often done through online A/B experimentation, which can be incredibly complex and time-consuming, especially considering the non-linear effects of these parameters on the metrics of interest.
In this tutorial, we will talk about how we can apply Bayesian Optimization techniques to obtain the parameters for such complex online systems in order to balance the competing metrics. First, we will provide an in-depth introduction to Bayesian Optimization, covering some of the basics as well as the recent advances in the field. Second, we will talk about how to formulate a real-world recommender system problem as a black-box optimization problem that can be solved via Bayesian Optimization. We will focus on a few key problems such as newsfeed ranking, people recommendations, job recommendations, etc. Third, we will talk about the architecture of the solution and how we are able to deploy it for large-scale systems. Finally, we will discuss the extensions and some of the future directions in this domain.Other authorsSee publication -
Online Tuning of Large Scale Recommendation Systems
Microsoft Machine Learning and Data Science Conference
Online tuning of recommendation systems can be incredibly complex. In this talk we will be looking at 2 systems (Notifications, PYMK-People You May Know). Models used in any recommendation system balance between multiple objectives like clicks, viral actions, sessions, disables. Constructing a serving algorithm to balance these objectives can be tricky and time consuming if there are more than 2 objectives. In addition, determining the parameters of the serving algorithm offline for…
Online tuning of recommendation systems can be incredibly complex. In this talk we will be looking at 2 systems (Notifications, PYMK-People You May Know). Models used in any recommendation system balance between multiple objectives like clicks, viral actions, sessions, disables. Constructing a serving algorithm to balance these objectives can be tricky and time consuming if there are more than 2 objectives. In addition, determining the parameters of the serving algorithm offline for non-stationary systems may not be possible. We model the objective and the constraints of the multi-objective optimization using Gaussian Processes. We will talk about how this technique is applied at scale and is used to serve Notifications to the members and People You May Know recommendations on the LinkedIn platform.
Other authors -
Near real-time optimization of activity-based notifications
KDD 2018
In recent years, social media applications (e.g., Facebook, LinkedIn) have created mobile applications (apps) to give their members instant and real-time access from anywhere. To keep members informed and drive timely engagement, these mobile apps send event notifications. However, sending notifications for every possible event would result in too many notifications which would in turn annoy members and create a poor member experience.
In this paper, we present our strategy of optimizing…In recent years, social media applications (e.g., Facebook, LinkedIn) have created mobile applications (apps) to give their members instant and real-time access from anywhere. To keep members informed and drive timely engagement, these mobile apps send event notifications. However, sending notifications for every possible event would result in too many notifications which would in turn annoy members and create a poor member experience.
In this paper, we present our strategy of optimizing notifications to balance various utilities (e.g., engagement, send volume) by formulating the problem using constrained optimization. To guarantee freshness of notifications, we implement the solution in a stream computing system in which we make multi-channel send decisions in near real-time. Through online A/B test results, we show the effectiveness of our proposed approach on tens of millions of members.Other authorsSee publication -
Fine-Grained Access to Online Content for Virtual Communities
The International Journal of Knowledge, Culture and Change Management
The specific knowledge-based technology that we are developing is Purple MediaWiki (PMWX), an extension of MediaWiki that supports fine-grained addressability of a knowledge repository. Unlike other web pages, content on a wiki is the result of a collaboration among the users of the wiki. As a result, the content on a wiki changes more frequently than most web pages. As more and more people add content to a web page and then refer to that content, it becomes important to pinpoint the location…
The specific knowledge-based technology that we are developing is Purple MediaWiki (PMWX), an extension of MediaWiki that supports fine-grained addressability of a knowledge repository. Unlike other web pages, content on a wiki is the result of a collaboration among the users of the wiki. As a result, the content on a wiki changes more frequently than most web pages. As more and more people add content to a web page and then refer to that content, it becomes important to pinpoint the location of the data for future reference or to provide a reference to someone else. The bookmarking option in a web browser allows one to bookmark the URL, but if this URL is the page as a whole it may be difficult for a user to locate the intended content when the amount of content on the page is large. With PMWX, each unit of an article (which may be as small as a single entry in a table) can be individually annotated and discussed by several individuals in a coherent fashion that allows the community to naturally break out into interest groups that can concurrently focus on aspects of concern to the community without losing overall participation by the entire community. The use of knowledge-based technology such as PMWX for building virtual communities is still in its infancy, but there are several communities that are using it on a regular basis. We will report on our experiences with the use of these technologies.
Other authorsSee publication
Patents
-
TUNING MODEL PARAMETERS TO OPTIMIZE ONLINE CONTENT
US https://www.patentguru.com/US2021089602A1
Courses
-
Convolutional Neural Networks for Visual Recognition
CS231
-
Deep Learning for Natural Language Processing
CS224d
-
Machine Learning
CS229
Recommendations received
5 people have recommended Viral
Join now to viewMore activity by Viral
-
Hi everyone! I'm designing an introductory machine learning class for junior and senior computer science students and need your help. For ML hiring…
Hi everyone! I'm designing an introductory machine learning class for junior and senior computer science students and need your help. For ML hiring…
Liked by Viral Gupta
-
Thrilled to share my recent interview with Economic Times on "Navigating the Evolving Landscape of Marketing." In this in-depth conversation, we…
Thrilled to share my recent interview with Economic Times on "Navigating the Evolving Landscape of Marketing." In this in-depth conversation, we…
Liked by Viral Gupta
-
The best software engineers are those who make the most efficient tools for others to be lazy, including themselves.
The best software engineers are those who make the most efficient tools for others to be lazy, including themselves.
Liked by Viral Gupta
-
Exciting insights in our latest blog! Dive into the intricacies of LinkedIn’s “People You May Know” (PYMK) technology as we tackle generating a…
Exciting insights in our latest blog! Dive into the intricacies of LinkedIn’s “People You May Know” (PYMK) technology as we tackle generating a…
Liked by Viral Gupta
-
I've been having a great time with Prometheus 2 over the last few weeks, and have a handy little notebook to share that should bootstrap your SFT…
I've been having a great time with Prometheus 2 over the last few weeks, and have a handy little notebook to share that should bootstrap your SFT…
Liked by Viral Gupta
Other similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore MoreOthers named Viral Gupta
-
Viral Gupta
Director at SPAN FOODS and a Blockchain & Cryptocurrency Visionary
-
Viral Gupta
UX Designer @ Domino’s 🍕 | Ex: Oracle
-
Viral Gupta
GSoC' 24 | 🚀 Vogue Virtuals | Web Developer | IOT | Video Editor
-
Viral Gupta
Looking traffic for Direct CPA,CPL,CPR and CPI Campaigns
16 others named Viral Gupta are on LinkedIn
See others named Viral Gupta