Deepak Saini

Deepak Saini

sainideepak119@gmail.com, desaini@microsoft.com

Hi there! I am Deepak Saini. I am a Research Engineer at Microsoft Research where I work on large scale recommendation problems. My work has an intersection between machine learning and efficient system design. I work on developing and deploying Extreme Classification (XC) algorithms with guidance from Manik Varma. Algorithms developed by me are deployed on Bing, MSN and other Microsoft products where they are helping hundreds of millions of users and (in-process) generating millions of dollars in revenue by showing more relevant and presonalized recommendations.
Before joining MSR, I worked as AI Engineer at Zomato to determine the best personalized ordering of restaurants for users on the App home page. Previously, I had obtained a Bachelor's and Master's degree in Computer Science and Engineering from IIT Delhi in 2018

News

  • Served as Session Chair for session on recommender systems at SIAM conference on data mining, Houston, 2024 April, 2024

  • Served as Program Committee member reviewing research papers for SIGIR Conference on Research and Development in Information Retrieval, Washington DC, 2024 January, 2024

  • Served as Program Committee member reviewing research papers for SIAM conference on data mining, Houston, 2024 November, 2023

  • Served as Program Committee member reviewing research papers for the Web Conf, Singapore, 2024 November, 2023

  • Won the Bing Ads Greatness Award in the Engineering Excellence category for work on optimizing online language encoders October, 2023

  • Served as Program Committee member reviewing research papers for WSDM Conference on Web Search and Data Mining, Mexico, 2024 September, 2023

  • Recieved Microsoft Patent Award for patent on OnlineGCN: Augmenting query representation with auxiliary information online for better retrieval September, 2023

  • Served as Program Committee member reviewing research papers for SIGIR-Asia Pacific Conference on Research and Development in Information Retrieval, Beijing, 2023 August, 2023

  • Deep Encoders with Auxiliary Parameters for Extreme Classification accepted at SIGKDD Conference on Knowledge Discovery and Data Mining, Long Beach, 2023 May, 2023

  • Personalized Retrieval over Millions of Items accepted at SIGIR Conference on Research and Development in Information Retrieval, Taiwan, 2023 April, 2023

  • Served as Program Committee member reviewing research and applied science track papers for SIGKDD Conference on Knowledge Discovery and Data Mining, California, 2023 March, 2023

  • Served as Program Committee member reviewing long and short papers for SIGIR Conference on Research and Development in Information Retrieval, Taiwan, 2023 February, 2023

  • Renee: End-to-end Training of Extreme Classification Models accepted at Conference on Machine Learning and Systems(MLSys), Miami, 2023 February, 2023

  • NGAME: Negative Mining-aware Mini-batching for Extreme Classification accepted at Web Search and Data Mining (WSDM) conference, Singapore, 2023 November, 2022

  • Recieved Microsoft Patent Award for patent on Personalizing per-Event Retrieval October, 2022

  • Now a Senior Applied Scientist at Bing Ads! August, 2022

  • Gave a talk on Extreme Classification: Retrieval Beyond Large Neural Language Representation Models at the Technical Advisory Board meeting (Microsoft Research India’s annual meeting) January, 2022

  • Gave talk at the 2021 Microsoft Research Summit in the "Future of Search and Recommendation" track October, 2021

  • Won the Bing Ads Greatness Award in the Engineering Excellence category for work on combining Extreme Classification and Siamese Networks October, 2021

  • SiameseXML: Siamese Networks meet Extreme Classifiers with 100M Labels accepted at International Conference on Machine Learning, Virtual Only, 2021 May, 2021

  • Recieved Microsoft Patent Award for patent on GalaXC (Graph neurAI networks with Labelwise Attention for eXtreme Classification) for Sponsored Search March, 2021

  • Gave a talk to the Financial Modelling Group at BlackRock on Extreme Classification and its potential applications in financial recommendations March, 2021

  • GalaXC: Graph Neural Networks with Labelwise Attention for Extreme Classification accepted at International World Wide Web Conference, Slovenia, 2021 January, 2021

  • Gave a talk on DeepXML: A Framework for Deep Extreme Classification Applied to Recommending 100 Million Items at the Technical Advisory Board meeting (Microsoft Research India’s annual meeting) January, 2021

  • DeepXML: A deep extreme multi-Label learning framework applied to short text documents accepted at Web Search and Data Mining (WSDM), Israel, 2021 November, 2020

  • DECAF: Deep extreme classification with label features accepted at Web Search and Data Mining (WSDM), Israel, 2021 November, 2020

  • Now a Research Software Development Engineer II at MSR! November, 2019

  • Joined MSR India as Research Software Development Engineer! Decemner, 2018

  • Joined Zomato as AI Engineer! June, 2018

Publications/Posts/Articles