Product Manager

• Jul 2022 - Present

Product

• Jun 2020 - May 2022

- Building self-serve, no configuration capability for the Customer CARE Platform - Currently ideating a no-code ML platform (AI Studio) where users can label datasets, train a model, check accuracy reports and deploy with one-click without any ML specific technical knowledge - Building self-serve, no configuration capability for the Customer CARE Platform - Currently ideating a no-code ML platform (AI Studio) where users can label datasets, train a model, check accuracy reports and deploy with one-click without any ML specific technical knowledge

Product Design Engineer

• May 2019 - Jul 2019

Industrial Product Design -- Designed an ergonomic and cost effective armrest and headrest for HR-CT scan Industrial Product Design -- Designed an ergonomic and cost effective armrest and headrest for HR-CT scan

Data Scientist

• May 2018 - Jul 2018

Multivariate Time Series Modelling -The objective was to forecast Bill of Lading (BOL) transactions in a month using various deep learning techniques -The time series was stationarized with greater than 95% confidence on the basis of the Augmented Dickey-Fuller Test -Built an LSTM (Long-Short Term Memory) Model in Keras to predict the BOL count with an accuracy of 85.18% -Tuned the hyper-parameters by changing the activation function of the LSTM model which enhanced the results by… Show more Multivariate Time Series Modelling -The objective was to forecast Bill of Lading (BOL) transactions in a month using various deep learning techniques -The time series was stationarized with greater than 95% confidence on the basis of the Augmented Dickey-Fuller Test -Built an LSTM (Long-Short Term Memory) Model in Keras to predict the BOL count with an accuracy of 85.18% -Tuned the hyper-parameters by changing the activation function of the LSTM model which enhanced the results by 6.2% Show less Multivariate Time Series Modelling -The objective was to forecast Bill of Lading (BOL) transactions in a month using various deep learning techniques -The time series was stationarized with greater than 95% confidence on the basis of the Augmented Dickey-Fuller Test -Built an LSTM (Long-Short Term Memory) Model in Keras to predict the BOL count with an accuracy of 85.18% -Tuned the hyper-parameters by changing the activation function of the LSTM model which enhanced the results by… Show more Multivariate Time Series Modelling -The objective was to forecast Bill of Lading (BOL) transactions in a month using various deep learning techniques -The time series was stationarized with greater than 95% confidence on the basis of the Augmented Dickey-Fuller Test -Built an LSTM (Long-Short Term Memory) Model in Keras to predict the BOL count with an accuracy of 85.18% -Tuned the hyper-parameters by changing the activation function of the LSTM model which enhanced the results by 6.2% Show less

Product Manager

• Dec 2015 - Jan 2016

- Aim was to make users sign up for education courses along with improving high level KPIs - Analysed the Acquisition, Behaviour and Conversion data of the users using SQL and Python to identify the drop-off point at a specific stage in the funnel - Shaped up the hypothesis by carefully investigating the user behaviour at junctures - Successfully decreased the bounce rate by 20%, ramp up the customer conversion rate by 33% and a significant increase in the DAU/MAU - Aim was to make users sign up for education courses along with improving high level KPIs - Analysed the Acquisition, Behaviour and Conversion data of the users using SQL and Python to identify the drop-off point at a specific stage in the funnel - Shaped up the hypothesis by carefully investigating the user behaviour at junctures - Successfully decreased the bounce rate by 20%, ramp up the customer conversion rate by 33% and a significant increase in the DAU/MAU