Aarthi S. Email & Phone Number

United States

Emeryville, California, Us

●Leveraged Databricks to develop a 79% accurate NER model using Adam Optimizer, enabling automated generation of concise product descriptions for seamless display on sales receipts●Led a Data Migration Program and navigated complex challenges in migrating data from legacy platform like AS400 to SAP, leveraging ETL processes to significantly enhance data quality and accuracy to 95%●Designed an automated analytical dashboard using Python, to showcase missing receipts and transfers to and from the WMS, which significantly enhanced data accuracy in business processes from 56% to 93%●Developed and implemented ETL jobs to automate the generation of store orders into SAP, reducing manual upload time by over 5 hours/day

Emeryville, California, Us

▪ Leveraged Databricks to extract and transform data from AWS, performed Exploratory Data Analysis (EDA) to identify significant features, and developed a Prophet forecasting model to predict sales and shipment, resulting in a 23% enhancement in operational efficiency and cost reduction in inventory management.▪Led the successful implementation of a Product Information Management (PIM) system, resulting in a significant improvement in data quality and accuracy across the organization▪ Developed a Python based automation pipeline to extract and transform images, elevating image availability to 97%, thereby enhancing the customer experience on platforms like Instacart▪ Built a Data Quality Dashboard to monitor and improve store data quality, achieving a 30% reduction in manual data correction effort▪ Implemented a ResNet-152 CNN Model to extract brand names and descriptions from diverse images, achieving an accuracy of 83%

Dodgeville, Wi, Us

Twitter Automated Posting with Hashtag Intelligence- Designed and Implemented a model to increase traffic to the website by converting daily subscription emails into automated twitter posts- Implemented Hashtag intelligence in the twitter posts to increase visibility leading to a 7% increase in organic social trafficProduct Listing Ads (PLA) Dynamic Pricing- Spearheaded and developed a new pricing model using competitor prices to increase margin and drive more clicks for PLAs- Analyzed and optimized the product data feed using the model, leading to an increase in revenue by a gross margin of $1MTime Series Forecasting of Revenue- Developed an 87% accurate time series forecasting model using Deep AR to predict revenue based on web traffic metrics- Built a KPI dashboard to facilitate YoY revenue analysis of the obtained results to help influence strategic decision making and drive profitGeneration of Articles and Product Descriptions using NLP- Achieved a $100K reduction in manual labor cost by developing a GPT-2 Model to auto generate articles and product descriptions- Pre-processed and tokenized the training data for the model, to produce accurate result texts for each product typeAnalyzing SEO Backlinks and Creation of Sitemaps- Extracted and analyzed the number of backlinks gained/lost per keyword, resulting in an average improvement in ranks by at least 2 spots- Developed a model to create and automate XML Sitemaps using canonical URLs, which eliminated manual labor costs and time by 90%Creation and Analysis of Product data feeds- Created and Automated Google PLAs, Facebook and Pinterest data feeds to update prices and URLs based on promotional changes- Performed statistical analysis on the data feeds which improved PLA and SERP rankings, resulting in increased traffic rate by 30%

Dodgeville, Wi, Us

Customer sentiment analysis using Facebook comments- Extracted user names, comments and reactions from posts on Lands’ End Facebook page using the Facebook API- Performed sentiment analysis on the user comments to identify the likes and dislikes of the user for a particular product- Created custom audiences based on the positive and negative reactions of the users to target personalised adsIncreasing New Customer Acquisition (NCA)Ad-Hoc Reporting/Projects- Personalized the index pages using a new sort order algorithm, which resulted in an 8% uplift on Purchases/User and Revenue/User

Champaign, Il, Us

Records and Information Management Services - Scraped text from the Government of Illinois email dataset and parsed it in order to feed them to processing tools - Built a predictive model using Bayes-classification to detect and obscure Personally Identifiable Information (PII) like SSN, phone number. The model had an accuracy of 93% in identifying PII

Espoo, Southern Finland, Fi

Access Management System (AMS) - Nokia Access Virtualizer Integration (NAV) - Developed a distributed state machine spanning across subsystems to monitor and notify the counterpart on state changes - Achieved 99.9% system uptime by implementing a proprietary extension to the NETCONF Framework that used the state machine- Enabled rich data-type support and extensibility by enhancing the interprocess messaging framework using YANG models

• Developed a video sharing portal for dubsmash videos built with PHP + JQuery + Mysql as backend and HTML,CSS for the front end

Master'S Degree, Information Management (Data Science And Analytics)

Bachelor Of Engineering (Be), Computer Engineering