Chaitra Hegde

I'm a alumni of NYU Center for Data Science, part of Courant Institute of Mathematical Sciences. Natural Language Processing and Computer Vision are my area of expertise.

Currently, I'm working as a Data Scientist at Fidelity Investments on NLP problems pertaining to customer interaction with Fidelity. As a member of Razavian Lab, I am working with Prof. Narges Razavian and Prof. Yvonne Lui on building a brain MR segmentation tool using deep learning models. You can contact me via email: cvh255[at]nyu.edu. Please have a look at my Resume for more details.

Education

  • New York University - Center for Data Science (Courant)

    MS in Data Science (Sep 2017 - May 2019)

  • Bangalore Institute of Technology

    Batchelor of Engineering in Computer Science (Jun 2013 - Jun 2017)

Research/Projects

  • Complete Anatomical Brain MR Segmentation Github

    We aim to build an alternative tool to Free Surfer (FS) that performs complete brain segmentation quickly and more accurately than FS [Extended Abstract]

  • Muscolo-skeletal MR Segmentation using dilated convolutions Github

    Used Convolution segmentation models to peform knee cartilage segmentation [Project Report]

  • Neural Machine Translation Github

    Bi-LSTM with attention and self-attention based models to perform Chinese to English and Vietnamese to English Translation

  • Yelp Recommendation Engine Github

    Built a SGD-based matrix factorization model to recommend restaurants to Yelp users [Project Report]

  • Bank Marketing Campaign Analysis Github

    Analyzed the prior marketing campaign of a Portuguese Bank using various ML models and recommend ways to improve the campaign [Project Report]

    •

    Work Experience

  • Fidelity Investments, Boston, USA

    Data Scientist (Jul 2019 - Present)
  • Working on NLP data from customer interaction with Fidelity and building NLP models to improve the customer experience
  • Up-to-date on NLP research

  • Bombora Inc., NYC, USA

    Data Science Intern (Sep 2018 - May 2019)
  • Built a Named Entity Recognition(NER) model using BERT, which is going into production. Reduced the cost of model by 30 times compared to Google Auto ML implementation of NER.
  • Built a 5800 Classification model using Self Attention, which is also going into production soon. Achieved 24% accuracy gain over the model in use.