Digital Marketing Data Analyst

Priyansh Agarwal

***–***-**** adbo7l@r.postjobfree.com linkedin.com/in/priyansh91/ priyanshagarwal.github.io SUMMARY

Analytical and innovative data science graduate with 3 years of professional experience working for fortune 500 companies, looking forward to providing solutions to the challenging problems faced in the field of Data Sciences and Machine learning. Expertise in Python, Business Development, Statistics, Azure, and Deep Learning. EDUCATION

Michigan Technological University Houghton, MI MS in Data Science GPA 3.75 August 2019 Deep Learning Nanodegree Udacity January 2018

Rajasthan Technical University Jaipur, India BS in Computer Science GPA: 3.55 May 2014 TECHNICAL SKILLS

• Python • SQL • Big Data • R • Deep Learning • Sklearn

• Computer Vision • TensorFlow • Map Reduce • Spark • Predictive Analytics • Inferential Statistics

• Azure • Keras • Artificial Intelligence • NLP • Machine Learning • Descriptive Statistics INDUSTRY EXPERIENCE

Enterprise Data Science Intern Kimberly-Clark Neenah, WI May 2018 – August 2019

Baby Connect – Explored and visualized daily activity data of 1000 infants to understand the daily cycle, food consumption habits and BM routines to predict diaper size and diaper consumption over time, yielded $3 million in customized marketing.

Modeled diaper changes over time to build a consumption curve for a customized replenishment model

Developed a diaper sizing model to predict diaper size change and retain user using digital marketing

Huggies Measurement – Engineered deep learning model to convert measurement pictures (average 1000 pictures daily) of a baby wearing diaper into data to study sagging of diaper overtime with pee events.

Position detection model to detect the position of the bay wearing K-C product, used transfer learning to build this model

Built object detect model using RCNN-Inception-V2 to detect different objects in a picture like diaper, study card and different marks on subject’s body for measurement, segment those objects and use Azure API to read study cards

Automated the whole process using Azure data factory and conserved $120K/Yr. in manual labor.

Gross Purchase Validator Bot - Used NLP, TF-IDF, and fuzzy logic to identify fraud address

Automated whole process using Azure data factory pipelines, blob storage, file system, and databricks notebook

Yielded 80K in saving’s per annum in manual testing, scrapped Google Maps, Facebook and Yelp to collect data.

Opportunity Gap Analyzer - Spearheaded Walmart and Kimberly-Clark marketing team to find an expected market share of Huggies around 60 Walmart CMA’s. Entire software was built in Python and visualized on Tableau.

Performed elastic net regression on selected features to find an expected market share of each product.

Resume Parser - Developed a keyword-matching algorithm in python to solve the complex problem of ranking resumes

Txt file of keywords is provided by the hiring manager, Used Word2Vec and Nearest Neighbor algorithms to rank resumes

Stored the results in Azure blob storage and presented those results in Power BI to various hiring managers.

Amazon Forecast Simulator – Build a price elasticity simulator to forecast 26 weeks of K-C products sold on Amazon.com.

Used Facebook’s Prophet to forecast sales for 80 products with more than 55 variables. Visualized those results in Tableau

Collaborated with business managers to enhance the accuracy of forecast at UPC level and eliminate manual intervention. Research Assistant Michigan Technological University Houghton, MI January 2018 - January 2019

Transferability in Artificial Intelligence (“The Unreasonable Ineptitude of Deep Image Classification Networks”)

The objective of this project is to figure out how the AI system would react towards transformed images and contours. I built the whole system on AWS deep learning AMI to trained Google's inception on 10 classes of birds and 10 classes of tools.

Extracted the weights for each transformation and plotted ROC curve for each transformation, which explained how the system was affected by different transformations, vision system was impaired by these transformed Images and had an accuracy of less than 30%.

Data Analyst WebCraft IT India Nov 2016 - June 2017

Customer Churn – Used machine learning algorithms to perform customer churn on e-commerce data.

The goal was to find important variables and reasons causing customer churn wrote an algorithm to compute churn score so that the marketing team can productively segment customers on the basis of churn score. Mobile Application Developer CG Technosoft India Jan 2015 – Nov 2016

Klik Night Life (Miami, USA): First of its kind in-venue social networking application for nightlife, over 10,000 downloads ACADEMIC PROJECT September 2017 - April 2018

Analyze IMDB Movie Reviews (Natural Language Processing - Sentimental Analysis)

Built an LSTM network to analyze 25,000 IMDB movie comment. Used NLP techniques to convert text into numbers.

Determine whether the movie review has a positive or negative sentiment, visualized the whole corpus using T-SNE graph. HOBBIES – Prospective pastry chef and ballroom dance enthusiast

Contact this candidate