Mar 2023 – Jun 2023 · 3 mos

Sep 2021 – Feb 2023 · 1 yr 5 mos

Jul 2020 – Sep 2021 · 1 yr 2 mos · Pune, Maharashtra, India

• Worked on developing python machine learning libraries to solve the problems of anomaly detection, time series classification and time series regression in an end-to-end manner.

Aug 2019 – May 2020 · 9 mos · Department of Mathematics

• To predict the 3D conformation from 1D protein sequence which can potentially revolutionalize the field of drug discovery and Bio-engineering as a whole.
• Re-implemented and compared, protein contact map and torsional angle based deep learning techniques. Proteinnet-12 dataset based on CASP-12 was used to reduce the data pre-processing time.
• Developed a neural network architecture which combines convolutional neural networks and BiLSTM based recurrent neural networks to assess the problem. And achieved good enough results to support the development of neural networks that learns to predict the 3D protein structure in an End-to-End manner.

May 2019 – Jul 2019 · 2 mos · Mumbai

• AI V/S TUBERCULOSIS
• To determine the factors affecting number of TB cases in a district in India and To better plan interventions for Adherence.
• Employed various tree based regression models to predict number of TB cases and decide most contributing features.
• Modelled 99DOTS callrecords as a HMM to determine the adherence pattern of an individual to the TB medication.
• MOBILE KUNJI
• This project was done in collaboration with BBC Media Action, and was funded by GATES Foundation.
• To assess the efficacy of “Mobile Kunji” and tailorit based on field worker’s usage patterns.
• Analyzed field worker’s callrecords and BBC call centerrecords to classify among unregistered mobile kunji’s users.
• Engineered various features from callrecords to assess the above classification and achieved max accuracy of 75%.