Sep 2023 – Oct 2025 · 2 yrs 1 mo · Bengaluru, Karnataka, India · Remote

• Led a project to create industry and customer segments from product usage data, and aid in
• designing marketing campaigns to target relevant customers for each product, leading to an
• increase in business revenue by more than 40%.
• Led a team of data scientists and data engineers to develop a hybrid ML and Gen AI
• framework to automatically create end-to-end credit proposal reports using internal and
• external data, increasing efficiency by more than 80%.
• Played a leading role in a project to summarize and identify relevant business opportunities
• from call reports using Gen AI, helping in revenue growth by more than 30%.

Aug 2021 – Sep 2023 · 2 yrs 1 mo · Bengaluru, Karnataka, India · Remote

• Led a team of data scientists and data engineers to build a recommendation system to make personalized recommendations of Zoom Apps to users, driving up installations by 20%.
• Played the role of the leading data scientist in a data team focused on creating business KPI’s, designing A/B tests and performing statistical tests to provide insights for gauging success of several experiments and help drive the launch of product features like Apps Bundle, leading to increase in Zoom Apps monthly Usage by 100%.
• Worked in a leading role in standardizing data models and extract features from product usage data to build a churn prediction model, which is expected to reduce revenue churn by an initial 20%.

May 2021 – Aug 2021 · 3 mos · Bengaluru, Karnataka, India · Remote

• • Predicted probability of customers to opt for a business product based on their information with an overall accuracy of around 85%.

Sep 2020 – May 2021 · 8 mos

• Created a statistical model to estimate cost on the basis of resource usage by customers for running a number of queries with varying degrees of complexity. A constrained polynomial regression model was built for the purpose and achieved an overall accuracy of 85%.
• Imputed data in time series containing a lot of extreme observations using a bi-directional LSTM with an overall accuracy of 80%, offering an improvement of nearly 40% on the existing techniques like Kalman Smoothing, Matrix Factorisation etc.
• Worked on a text recognition model to identify inverted texts in images with an overall accuracy of 90%.

Aug 2018 – Sep 2020 · 2 yrs 1 mo

• Forecasted orders for China Walmart to aid in efficient workforce management using a number of classical and deep learning based models like Regression ARIMA, Bayesian Structural Time Series, Neutral Net etc. with an innovative additional adjustment made to account for abrupt upward fluctuations in the data. An overall accuracy of 95% was achieved across 600 different stores with roughly 100k orders being placed on a daily basis.
• Analysed the effect of Gaussian Noise on the accuracy of deep learning based forecasting models and quantified how addition of noise actually helps in building a better neural network model.
• Created a robust measure of forecast accuracy to quantify performance of time series models much more appropriately.
• Calculated embeddings for text data on the basis of contexts in order to facilitate clustering of said data and help gauge customer sentiment better.
• Used PySpark to create modules to automatically pre-process data and perform a number of EDA operations including statistical tests and help users gain a detailed insight of the data.
• Extracted latent variables from categorical data to facilitate clustering of such data in order to obtain better models.

Jan 2018 – Jun 2018 · 5 mos · Pune Area, India · On-site

• Predicted NPS scores using relevant ML models like Random Forests, XGBoost, MARS and SVR to gauge customer feedback better.
• Worked on extracting and visually representing critical information from a drivers' health survey to help create improved safety guidelines and suggestions for a better lifestyle for them.
• Forecasted regional water level using time series modeling techniques like ARIMA, TBATS, neural network, theta forecasting etc. with an overall 82% accuracy as a part of a Corporate Social Responsibility project to help the concerned authorities plan accordingly.

May 2017 – Jul 2017 · 2 mos · Bengaluru, Karnataka, India · On-site

• Built a Vector Autoregressive Model to forecast interconnected economic factors like GDP, HPI and Unemployment Rate.
• Created a cycle index based on the forecasted values from the VAR model, using HP Filter and Principal Component Analysis, to predict potential economic stress scenarios.

Sep 2015 – Jul 2016 · 10 mos · Kolkata Area, India · On-site

• Worked on creating a mechanism to migrate data from Oracle to SAP databases using PL-SQL.