Lead - Electoral Analysis • Lead a team of 4 as part of the electoral strategy team at IPAC HQ. • Conducted post-election analysis of Bihar State Legislative Assembly elections 2020 - impact of third alliance, trend changes through the phases, major vote banks, and variations from previous election results.• Electoral requirements analysis for West Bengal and Tamilnadu - historical election results, impact of leaders jumping ship, and safe winning margins, resulting in the cementing of winning alliance in Tamilnadu.Lead - Tamilnadu Campaign Data Team Telephony Unit• Lead a team of 6 - the data acquisition team vis-a-vis call centre, IVR and SMS, and directly supervised the call centre - 400+ strong.• Re-structured the operations, divided responsibilities, identified KPIs for hourly and daily performance monitoring, streamlined the training process, gamified bonus and incentives, established communication channel from the calling agents till the upper management, increasing the operational efficiency by 56%, accuracy by 600% and decreased attrition by 44%

Product Recommendation System • Built a user and product based collaborative filtering system to integrate into a targeted marketing strategy which boosted sales conversion by 27%.

PRODUCT RECOMMENDATION SYSTEM• Product Recommendation system - Built a hybrid user-based and item-based collaborative filtering system based on kNN criteria on clickstream data, thus fast tracking the purchasing process and increasing purchasing rate by 4%.

• Online funnel dropout analysis - Developed a Markov chain implementation of attribution model to capture customers' website interaction to trace the checkout and dropout paths, conducted A/B tests, which improved positive funnel transition rate by 12%.

• Identifying bill related calls - Built a classifier based on historic monthly bills for customers to identify repeat callers and performed a fishbone RCA to investigate the reason followed by cost benefit analysis, saving $30M per year.

• Multi-channel customer interaction pattern identification - Employed PrefixSpan to capture customer online touchpoint patterns that led to customer care calls, which led to timely intervention and registered a drop of 3% calls, saving $1.2M per month.

SPATIAL ANALYTICS – Identify the best site for a retail outlet in a given region.• Plotted all civil structures on a map using GPS information using ArcGIS.• Built a Huff Gravity model using GPS and demographics data to calculate average footfall across the map.• Predicted sales across the map using inventory data Quantitative Outcome – Located site with potentially 14% more footfall and 9% more sales than other shops.

• Deployed an end-to-end automated solution pipeline to predict suitability of 500K+ candidates to 35 different job profiles - which resulted in an increase in candidate placement rate by 7% to 8%.

DESIGN OF MULTI-FUNCTIONAL HOSPITAL BED• Designed and built a multifunctional hospital bed with Trendelenberg function, sit-up/recline modification facility, bed height adjustment, completely using mechanical components, operated by pedals.Quantitative Outcome – Capped the maximum operation effort to 5kg and reduced turns per oscillation by 50%.

TRAVEL FARE REVISIONo Revisal of the then prevailing auto rickshaw travel fares in Chennai, through survey and analysis, which was later presented to the Government of Tamilnadu resulting in the implementation of metered auto scheme in the state.