Automation of Fault detection in user techniques • The client was taking a step towards smart fitness equipment’s with sensors that can detect and correct flaws in postures and movements while exercising.• A test study was conducted for close to ~1000 people with age groups, height, weight, gender, and other relevant attributes, to understand pattens of good/bad user techniques.• Dynamic Time Warping was used to calculate the similarity of good and bad pattern,• The algorithm… Show more Automation of Fault detection in user techniques • The client was taking a step towards smart fitness equipment’s with sensors that can detect and correct flaws in postures and movements while exercising.• A test study was conducted for close to ~1000 people with age groups, height, weight, gender, and other relevant attributes, to understand pattens of good/bad user techniques.• Dynamic Time Warping was used to calculate the similarity of good and bad pattern,• The algorithm was deployed on the cloud and was connected with the real time sensors of the equipment, which gave real time feedback to the users.Creation of automated trading signals for the traders• Automated reports using indicators (EMA, RSI, Stochastic etc.) as well as price patterns.• Built a machine learning algorithm using dynamic time warping to detect patterns like head & shoulders, channels etc. on a real time basis across 40+ instruments. • All the reports were part of the real-time automated dashboard, along with emails alerts for important buy/sell signals.Creation of end-to-end process flow for implementation of the trading strategy • Created a 6-step process flow to for optimization and implementation strategies:o Basic idea generationo Updating of the idea through extensive business research and understanding o Back testing and optimization of the idea with iterations o Construction of a layer of deep learning neural network models for risk mitigations and profit optimizationso Testing of the idea in the Demo environment for validation, o Choose the final list of combinations to Go Live if we get a go ahead from all the above points. Show less

Lead the delivery of Multiple Use cases for a Telcom giant in Africa & the Middle East• Customer 360 o Integrated analytics data marto Creation of Unique Global Customer IDo Multi-dimensional segmentation engineo Customer Profitability and Lifetime Value Models• Customer Value Management o Customer Growtho Churn Management Programo Campaign strategy and offer design.o Cross-sell/Up-sell• Smartphone Penetration - Recommendation engine for Smartphone… Show more Lead the delivery of Multiple Use cases for a Telcom giant in Africa & the Middle East• Customer 360 o Integrated analytics data marto Creation of Unique Global Customer IDo Multi-dimensional segmentation engineo Customer Profitability and Lifetime Value Models• Customer Value Management o Customer Growtho Churn Management Programo Campaign strategy and offer design.o Cross-sell/Up-sell• Smartphone Penetration - Recommendation engine for Smartphone upgrade• Tarif Rationalization o Right-size Tariff rationalization engineo Tariff migration matrix at customer level Show less

Built a product that forecast sales of phones using social/digital data.• Built several machine learning models to estimate the sales of the phones in the Launch/In life phase using historical sales, promotions as well as digital data.Understand the impact of Paid Ads on Organic clicks• Built several machine learning models to estimate the cannibalization of Organic clicks due to paid Ads.Estimation of ROI and effectiveness of media (Television Network)• Built a MCMC model… Show more Built a product that forecast sales of phones using social/digital data.• Built several machine learning models to estimate the sales of the phones in the Launch/In life phase using historical sales, promotions as well as digital data.Understand the impact of Paid Ads on Organic clicks• Built several machine learning models to estimate the cannibalization of Organic clicks due to paid Ads.Estimation of ROI and effectiveness of media (Television Network)• Built a MCMC model to estimate the ROI of various marketing channels for different parts of India.• Optimal level of Promotions for various promotion Vehicle was estimated which in turn helped client increase/decrease their existing spends. Show less

Built an Activation Model• Built several machine learning models for a Technology client to find out the most likely customers to take the subscription. • The project also helped the client understand the key drivers of subscription.

Measuring Impact of Campaign for pharmacy client retail chainBusiness Problem• A major pharmacy retail chain based in the United States wanted to understand the effectiveness of their monthly campaigns Analytical Solution• An automated script in SQL was written to fetch data from Teradata, Incremental efficiency was calculated using test and control methodology• A standard report was made, which was generated every month, week, quarter etc. These reports provided… Show more Measuring Impact of Campaign for pharmacy client retail chainBusiness Problem• A major pharmacy retail chain based in the United States wanted to understand the effectiveness of their monthly campaigns Analytical Solution• An automated script in SQL was written to fetch data from Teradata, Incremental efficiency was calculated using test and control methodology• A standard report was made, which was generated every month, week, quarter etc. These reports provided recommendations to improve campaign success and drive higher sales among the loyalty members. Impact.• The reports generated insights into customer behaviour and spending to augment better targeting and boost sales.Route Optimization for a convenient store Business Problem• The largest retail convenience stores chain in the US wanted to optimize their route planning to minimize the number of trucks and distance covered for optimal supply chain management• Client had to replenish several stores from DC dailyAnalytical Solution• Automated system was built that used web scrapping to get all possible distance and timing from stores to DC and within stores • A customized algorithm based on clustering of stores was developed for optimal used of trucks and distance Show less