Sayali Ghodekar

Sayali Ghodekar

Data Scientist

DeepAffects

About Me

Hello!

I am a Data Scientist working at DeepAffects. DeepAffects is an AI powered Voice Intelligence startup where we work on Automated multi-speaker recognition, voiceprints, emotions & intent extraction from natural conversations. Prior to that I was a research intern at the Center For Indian Language Technology at IIT Bombay, under the guidance of Dr. Pushpak Bhattacharyya and Dr. Malhar Kulkarni. Here I worked on distributed representations for Indian Languages and their applications for Automated Cognate Detection and Computational Phylogenetics. I also had an opportunity to work in the exciting field of Cognitive NLP.

In my free time I love reading books and write articles. I am a part of the vibrant reading community Readers By The Bay and often contribute to Coffee Writers Blog.

Interests

  • Natural Language Processing
  • Computational Linguistics
  • Speech Processing
  • Psycholinguistics

Education

  • Bachelor of Engineering in Computer Science, 2019

    Pune Institute of Computer Technology, Pune University

Experience

 
 
 
 
 

Data Scientist

DeepAffects

Jan 2020 – Present Mumbai

Responsibilities include:

  • Research and analysis of NLU and speech intelligence of multi-speaker conversations.
  • Development and scaling of DeepAffect’s abstractive summarization API which processed over 1M+ minutes of audio into summaries.
  • Built DeepAffect’s conversation analytics and metrics including dialogue act tagging, intent classification, question and answering systems using PyTorch.
 
 
 
 
 

Research Intern

Center for Indian Language Technology, IIT Bombay

Jun 2019 – Dec 2019 Mumbai

Responsibilities include:

  • Research in the distributed representations for Indian languages.
  • Investigations into the cross-lingual embeddings and transfer for low-resource NLP.
  • Developed the novel Textual History Analysis Tool to capture the historical evolution of texts through various temporal stages and digitized manuscripts using computational phylogenetics.

Contact