Mukund Rungta

As an Applied Scientist at Microsoft in Cambridge, MA, I specialize in applied research within Natural Language Processing (NLP). I'm an part of the Micrsoft AI Development Program (MAIDAP), contributing to cutting-edge advancements in AI/ML at Microsoft.

I pursued my Master's degree in Computer Science with specialization in Machine Learning (Thesis) at Georgia Tech. Under the supervision of Prof. Chao Zhang, I completed my thesis, further honing my expertise in NLP. Additionally, I had the privilege of collaborating with Prof. Diyi Yang, exploring the intersection of Social Computing and NLP, broadening my understanding of these interdisciplinary fields.

Prior to this, I graduated from IIT Delhi in 2018 with a major in Computer Science. My bachelor thesis was on deep learning methods for animal detection in low-memory setting under Prof. M. Balakrishnan and Prof. Chetan Arora.

Now, as a member of Microsoft's MAIDAP team, I'm dedicated to pushing the boundaries of NLP, leveraging my diverse academic background to drive innovation and solve real-world challenges.

Email  /  Resume  /  Google Scholar  /  Twitter  /  Linkedin

profile photo
Experiences
project_img

Applied Scientist at Microsoft, Cambridge MA

  • Part of Office of CTO at Microsoft
project_img

Software Engineer-2 at Microsoft, India

  • Responsible for end-to-end ownership of building and deploying machine learning model for classification of verbatims by Admin Users of O365 products.
  • Worked towards automation of monthly reports for analyzing the Net Promoter Score(NPS) change across various segments and dimensions.
  • Integrated Logging & Incident management service with the Service Fabric application for application management in production.
project_img

Machine Learning Engineer at Samsung, Bangalore

  • Developed and commercialized On-Device model for transliteration of English to Korean & vice versa for relevant search results for name of person and geographical locations.
  • Modelled Transformer based multi-head attention for sequence labeling with pre-trained embedding to extract important key phrases from text documents like messages, emails and notes.
  • Integrated the search with Contacts Provider viz. Dialer, Call Log, Contacts by implementing the Boolean search queries. Improved the performance of Contact Search by 20X by using on demand filling of Cursor Window
  • Commercialized end to end algorithm and flow for suggestion of folder names based on its constituent applications. Released as plugin in latest Home Star - OneUI2.0
Publications

(Design & CSS courtesy: Jon Barron)