Pooja Gurav

Graduate Data Science student with 3 years of professional experience in the Information Technology industry, specializing in data analytics and software development. Skilled in leveraging data-driven insights to support business decision-making.

About

Projects

Stock Market Prediction using LSTM

Developed an LSTM-based time-series model for Tesla stock price prediction, reducing testing RMSE by 30% compared to Logistic Regression and XGBoost. Conducted EDA, feature engineering, and created interactive dashboards with visualizations (line charts, heatmaps, AUC-ROC) to improve model.

Predicting county-level poverty rates across the United States- RA Project

Developed a sentiment analysis system using deep learning, achieving 72.81% adjusted R² and 80% accuracy in poverty analysis. Used Pandas and R for preprocessing, modeling, and validation. Enhanced precision through probability threshold optimization.

Credit Card Fraud Detection using Machine Learning

Built and evaluated ML models (KNN, Logistic Regression, SVM, Decision Tree) on a credit card fraud dataset with 284,808 transactions. Used PCA for dimensionality reduction, SMOTE for class imbalance, and cross-validation for tuning. KNN and Decision Tree achieved over 90% accuracy.

Bone Fracture Detection using Deep Learning

Developed and fine-tuned deep learning models, including a custom CNN and ResNet-18, for bone fracture detection from X-ray images, achieving 80.1% accuracy through transfer learning. Analyzed model performance and implemented strategies to mitigate underfitting, overfitting, and dataset imbalance, showcasing the potential of AI to improve diagnostic accuracy and efficiency in medical imaging.

821f6a4e-ad6f-41ab-91e8-07793f1f8b93_20240301_Corporate_application_header_seatbelt_detect

Distracted Drivers Detection

Built a real-time distracted driver detection system using CNNs with TensorFlow, Keras, and OpenCV on 22K+ images. Enhanced accuracy via image preprocessing and visualized data with FiftyOne. Optimized for edge deployment, enabling in-car feedback by classifying driver behavior into 10 categories.

RecruitMatic: AI-Driven Talent Discovery, Simplified

Created 'RecruitMatic,' an AI-powered platform for recruiters using OpenAI API, Python, and Streamlit, boosting performance by 30% and document parsing by 35%. Designed an intuitive UI to streamline workflows, enhance candidate insights, and improve data extraction accuracy, reducing response times.

Skills

I am well-versed in various programming languages and data analysis tools, with hands-on experience in working with statistical methods, predictive modeling, and large language models. I have applied techniques like regression analysis, data mining, and feature engineering in real-world projects. Additionally, I am familiar with cloud platforms, visualization tools, and collaborative development environments, enabling me to efficiently analyze, interpret, and communicate data-driven insights.

Data Analysis Techniques

Regression Analysis, Big Data Analytics, Data Mining, Predictive Modeling, Generative AI, Large Language Models, Data Cleaning and Wrangling (Pandas, Numpy), Data Visualization, Statistical Analysis, Feature Engineering

Programming Languages

C++, C, Python (Proficient), R (Proficient), SQL (Proficient), HTML(Intermediate), AMPL, MySQL

Software and Tools

AWS, Google Colab, Advanced Excel, GitHub, Tableau, Power BI, Databricks, Microsoft Office suite, Azure, Git, Google Cloud Platform (GCP)

Experience

University of Michigan- Dearborn

Graduate Research Assistant

As a Research Assistant, I led the design and implementation of various linear and logistic regression models in R to predict county-level poverty rates nationwide. I performed exploratory data analysis (EDA), feature engineering, and data preprocessing to identify the primary socio-economic drivers of poverty. Through cross-validation and model tuning, I improved predictive performance, achieving a 72.81% adjusted R² for regression and 80% classification accuracy. Additionally, I developed data visualization dashboards and performance reports using Pandas, ggplot2, and Matplotlib to effectively communicate model results and actionable insights to stakeholders.

Accenture

Data Analyst

As a data analyst, I used Python and SQL to automate document reviews and workflow assessments, which almost tripled manual labor and allowed for data-driven process enhancements. In order to expedite reporting and analysis, I created Tableau dashboards coupled with document data and constructed end-to-end ETL pipelines using Python, SQL, and Azure Data Factory. I decreased process mistakes by 25% by putting machine learning models for forecasting and root cause analysis into practice using Scikit-learn, TensorFlow, and Statsmodels. Furthermore, I improved the accessibility and quality of the data by thoroughly cleaning, validating, and integrating it with SQL and Azure Databricks, which greatly increased the dependability and speed of insights. In order to reduce manual labor and speed up data-driven decision-making, I also used Power BI and Excel to optimize more than 70 customer surveys and 40 company assessments.

Accenture

Application Development Analyst

I used Pandas and NumPy to clean and transform big datasets, which improved the quality of modeling inputs and reduced data inconsistencies by 32%. Using scikit-learn, I developed and refined machine learning models that enabled predictive insights to assist business choices and achieved up to 87% accuracy in categorization tests. In order to identify important trends, I also carried out exploratory data analysis (EDA) and produced visualizations, which enhanced reporting effectiveness and directed strategic activities.

Certifications

Complete Guide to Generative AI for Data Analysis and Data Science Complete Guide to Generative AI for Data Analysis and Data Science

AWS Educate Introduction to Generative AI - Training Badge

Power BI Essential Training

Google Cloud Data Engineering Foundations

Generative AI: Working with Large Language Models

SQL for Data Analysis

Complete Data structure and Algorithm Course in Python