Consumer-centric Data Scientist from the pale blue dot 🌏

Currently Data Scientist at Tiger Analytics | alum BITS Pilani

[email protected] | LinkedIn | Twitter | Resume

👀 Open to opportunities 👀

<aside> 💡 Hello! I am Harshit, an engineer turned data scientist, powered mostly by the drive to solve new problems - big or small. I drive collaboration to meet real user needs, delivering outcomes over output. My approach to solving real problems includes ML, Stats and loads of Business Context, with exposure to Insurance, Consumer Retail, Entertainment, Automotive and Pharma sectors.

</aside>

Education: **Birla Institute of Technology & Science, Pilani (**2012 - 2017) ****M.Sc. (Hons.) Biological Sciences, B.E. (Hons) Mechanical Engineering (Dual Degree)
Language Proficiency: Python, R, C++, SQL

📊 So far

Data scientist at SF-based — Tiger Analytics Jul ’17 - Present

Worked with Consumer Retail, Insurance and Media clients to solve Product Recommendations, New Customer Acquisition, Item Advertisement Attribution, and Customer and Product Segmentation.

Recommendation Engine – Led a small team of analysts in building a wine recommendation engine from scratch using a combination of collaborative filtering techniques, product associations, NLP tools, and multiple business rules. Delivered over $ 4 million in revenue through personalized email campaigns in the last year
Halo/Cannibalization and Cross-sell study - Built an attribution model to understand the incremental impact of advertising an item and help fine-tune it in order to increase customer footfall and boost the sales for homegrown wine brands having higher margins, leading to a shift in advertisement strategy.
Interpretable propensity modeling frameworks using a combination of tree-based (GBM, XGBoost) and regression (GLMNet and backward logistic) algorithms in python and R in order to

Identify the customers most likely to go through the marketing funnel
Reducing the customer acquisition cost to below 20$

Developed a customer segmentation framework and conducted several segmentation studies using partitioning, hierarchical and graph-based clustering methods, combined with apriori rules mining for forming market baskets -

Developed personas to assist targeted marketing campaigns
Understand consumer behavior and television consumption habits to address the discerning viewership
Segmented programs on TVR and viewership metrics to suggest the feasibility of a niche product offering