Professional Experience
Henkel
- Lead and collaborated with Supply Chain Customer Service team to improve expedited shipping transactions by applying classification models which provided inference into discretionary operating costs that can be reduced $1.2M annually.
- Partnered with Supply Chain Planning Network Development team to consolidate 1TB of Big data from multiple sources for reusability and KPI design through Power BI to identify underperforming worksites per region.
United Parcel Service (UPS)
- Collaborated with International Marketing team to research & manage an in-house Topic Model solution using BERTopic with LLM representation on 200GB of unstructured customer verbatim data reducing operational costs $15K annually.
- Worked with Competitive & Market Intelligence team to develop an in-house Generative AI application by accurately synthesizing 1TB of unstructured financial data & applying RAG techniques resulting in coherent answers to user questions.
- Analyzed optimization problem of 10 KPIs for Marketing team, identified non-stochastic constraints, and provided alternative strategies.
- Analyzed 15 KPIs monthly to Data Analytics & Business Enterprise Business Intelligence team providing insights on impacted delivery density for business use & contributed to script pipeline enhancing efficiency by 2 minutes.
Quest Diagnostics
- Developed PoC Deep Learning Negation & Entity Disambiguation model extracting pertinent information on 1TB of unstructured clinical data automatically populating structured databases using BERT & OpenNLP toolkits via PyTorch
- Applied NLP Text Sentiment analysis on 2GB of unstructured medical text to assess competitor markets for future strategies.
- Lead Time Series Analysis on 70GB of logistics data using XGBoost, formulating 10 KPIs that improved Turnaround Time.
- Engineered and executed ELT/ETL pipelines to process and centralize large-scale datasets, including migrating 10GB of client data using Informatica Cloud, transforming 250TB of clinical Big Data within Azure Cloud as a pre-stage for future application development, and integrating 320GB of unstructured data sources using Matillion to process into Snowflake to house data.
- Conducted Customer Segmentation on 500GB of demographic medical data using clustering and classification models to market diagnostic tests to potential consumers and drive sales. Communicated insights and next steps to stakeholders.
Data Science Projects & Contributions
Contributions
Matching Messy Pandas Columns
Resolved missing edge cases in fuzzy matching logic.
- Improved the accuracy of fuzzy matching for messy pandas columns.
- Collaborated with the author to ensure comprehensive coverage of all potential use cases.
Computer Vision
Face Detection Project
Performed face detection on AI-generated images.
- Developed classification models to differentiate real faces from AI-generated ones.
NLP Sentiment Analysis
Twitter Gillette Unigram Model
Built unigram sentiment models for analyzing Twitter campaigns.
- Extracted detailed insights into social responses to Gillette ads.
Amazon Collagen Reviews
Sentiment analysis of customer reviews on Amazon.
- Scraped unstructured text data from Amazon product reviews.
- Performed sentiment analysis using NLP models.
- Optimized models for better accuracy.
Twitter #GilletteAd
Analyzed public sentiment for the Gillette Ad campaign.
- Built models to classify and interpret tweet sentiments.
- Derived actionable insights for marketing teams.
Amazon Preworkout Reviews
Analyzed preworkout product reviews on Amazon using NLP.
- Extracted insights to improve customer experience.
Twitter #BoycottGillette
Explored Twitter sentiment for #BoycottGillette.
- Scraped tweets related to the #BoycottGillette campaign.
- Performed sentiment analysis and visualized trends over time.
- Developed insights into the public's reaction to the campaign.
Yelp Restaurant Reviews
Applied sentiment analysis using advanced NLP techniques.
- Scraped Yelp reviews for restaurant data.
- Derived insights to help restaurant owners understand customer satisfaction.
Classification & Regression Analysis
Predicting Hypertension & Visualizing Insights
Analyzed public health data to predict hypertension risks.
- Applied models like Logistic Regression and XGBoost.
- Provided feature importance analysis using XGBoost models.
- Provided actionable insights for healthcare professionals.
Predicting Used Vehicle Prices
Developed predictive models for used vehicle prices.
- Applied machine learning techniques including Random Forest and XGBoost.
- Engineered features to refine prediction accuracy.
Predicting Bank Loan Approval
Developed predictive models to assess bank loan approvals.
- Engineered features to improve model accuracy.
- Applied Random Forest and Logistic Regression models.
Predicting Diamond Prices
Predicted diamond prices using regression models.
- Engineered features to improve price prediction accuracy.
- Applied advanced regression techniques.
Predicting Home Prices
Analyzed real estate data to predict home prices accurately.
- Engineered key features to enhance model performance.
- Built regression models using Random Forest and XGBoost.
Skills