Data Scientist Science

Pranjali Deshmukh

862-***-**** *******************@*****.*** linkedin.com/in/pranjali-deshmukh/ github.com/PranjaliD11 Portfolio Experience

Data Scientist August 2024 –Present

CrowdDoing California, US

• Built and optimized an NLP-based sentiment classifier (Naive Bayes), automating analysis of 10,000+ feedback entries and improving recall by 10%

• Led data wrangling and feature engineering efforts on diverse datasets, enhancing model accuracy and reducing pre processing time by 25%.

• Improved sentiment analysis model accuracy by fine tuning and hyperparameter optimizing, resulting in 92% accuracy and 10% improvement in recall.

• Collaborated cross-functionally to develop Tableau dashboards for sentiment tracking, boosting responsiveness and stakeholder alignment.

• Automated the categorization and analysis of user feedback, enabling sentiment tracking and data-driven decision-making while reducing operational bottlenecks and improving responsiveness to user concerns. Data Science Intern May 2023–July 2023

Ashwa Enterprise LLP Maharashtra, IN

• Cleaned and preprocessed a dataset of 50,000+ customer records by handling missing values, outliers inconsistencies, hence improving data quality and usability for analysis.

• Conducted customer segmentation using K-Means clustering to refine marketing outreach, improving campaign responses.

• Performed exploratory data analysis(EDA) to identify trends, patterns and correlations in customer behavior, providing valuable business insights and improving decision-making.

• Engineered a decision tree model to forecast engagement levels with 85% accuracy using behavioral metrics.

• Created interactive Tableau dashboards to visualize key insights, leading to 20% increase in market campaign response rates Data Scientist March 2020 –July 2022

Binary Data Labs Maharashtra, IN

• Implemented an Extreme Gradient Boosting classification model to predict customer churn with 87% accuracy, leading to a retention strategy that reduced churn by 15% over six months.

• Collaborated with cross-functional product teams to refine predictive models, enhancing campaign effectiveness which contributed to a 10% increase in customer engagement metrics across multiple platforms.

• Analyzed consumer behavior data and identified key trends, resulting in the development of targeted marketing campaigns that decreased customer churn by 10% within a single quarter, impacting over 5,000 active users.

• Managed the extraction, cleaning and pre-processing of a 1M+ customer dataset by handling missing values, performing feature scaling and encoding categorical variables, to ensure high-quality data for model development.

• Applied SMOTE(Synthetic Minority Over-sampling Technique) to address class imbalance,increasing recall by 15% and improving model’s ability to identify more potential churn cases, enhancing retention efforts.

• Designed interactive Power BI dashboards to visualize risk levels and key performance indicators(KPIs), enabling stakeholders to take timely actions and optimize decision-making. Technical Skills

Programming Languages: Python, R, SQL, VBA, PL/SQL Machine Learning : TensorFlow, Scikit-learn, NLP, Regression, Classification Data Visualization: Tableau, Power BI, Seaborn, Matplotlib, Looker, Plotly, ggplot Statistical Analysis: A/B testing, ANOVA (analysis of variance), Chi-square test, Hypothesis testing Tools: GCP, SPSS, AWS, Azure, SSIS, ArcGIS, Jupyter notebook, Microsoft Excel, Notion Databases: Oracle, SQL Server, MongoDB, Snowflake, MySQL, PostgreSQL Education

New Jersey Institute of Technology New Jersey, US

Master of Science in Data Science, Statistics Track May 2024 University of Mumbai Maharashtra, IN

Bachelor of Engineering in Computer Science May 2022 Projects

AI vs Human image classification Python, VIT, Transformers, Hugging face Blog February 2025

• Fine-tuned and utilized Google’s Vision Transformer model to classify images.

• Successfully predicted over 11Gb of images as either AI generated or not. Depression Detection Python, Xgboost, Logistic Regression Github November 2024

• Handled missing values, inconsistencies in dataset,along with mapping to prepare data.

• Deployed and compared classification models, Cat-boost performed binary classification with 0.92.

Contact this candidate