Natasha Pashupathi profile picture

Hello, I'm

Natasha Pashupathi

Data Scientist

My LinkedIn profile My Github profile

Get To Know More

About Me

Professional Experience

4+ Years of Expertise
in Data Science & Software Development

Education icon

Education

PES University
Bachelor of Engineering in Computer Science
2015 - 2019

University of Maryland, College Park
Master of Science in Information Systems
2023 - 2024

I’m a data science enthusiast with over four years of experience turning data into actionable insights and building software solutions. With a Bachelor's in Computer Science and a Master’s in Information Systems from the University of Maryland underway, I combine data science expertise with a strong software engineering background. I’ve led projects across industries, leveraging machine learning, advanced analytics, and robust software development to drive business success. Passionate about innovation, I’m always learning and adapting in the ever-evolving world of data science and technology.

Professional Experience

Hughes
May 2024 - August 2024
College Park, MD
Data Science Intern
● Led automation of order placing using GCP and APIs reducing manual effort
● Built ETL pipelines with Google Dataflow and Apache Beam to process and load order data into AlloyDB
● Developed an API using Python Flask for order processing, increasing accuracy by 35%
● Implemented data storage and backup solutions using GCP BigQuery, improving efficiency by 40%
Bosch Global Services
Jul 2019 - Apr 2023
Bengaluru, India
Software Engineer
● Developed QML and C++ applications for premium BSH products integrating predictive models, improving power usage efficiency by 20%
● Designed a drag-and-drop UI tool in Qt Creator with data visualization features
● Created Python-based data pipelines and test automation frameworks, cutting bugs by 10%
Fidelity Investments
Jan 2019 - Mar 2019
Bengaluru, India
Data Science Intern
● Analyzed datasets with 88 features across investment funds
● Developed an XGBoost model using SHAP values for loss prediction with 89% accuracy and 77% recall
● Built an automated system to extract data from financial documents, improving efficiency by 25%
University of Maryland, College Park
Aug 2023 - May 2024
College Park, MD
Graduate Teaching Assistant
Assisted in teaching graduate-level data science courses, providing support in Python programming, and mentoring students on projects.

Browse My Recent

Projects

Project 1

Cross-lingual Question Answering System

Built a multilingual QA system by translating a Hindi dataset and using fuzzy matching for accurate answers. Optimized preprocessing with lemmatization and batching, cutting training time by 20%. Trained a BERT model with AdamW, achieving 90% accuracy, 78% F1, and 65% EM, and deployed a Gradio app for 1-second response time.

Project 2

Conversion of ASL Hand Gestures to Text and Speech in Multiple Languages

Built an accessible hand gesture recognition system that converts images into text and speech for users with disabilities. Leveraged Python, OpenCV, and LabelImg for image processing and labeling. Trained an SSD MobileNet V2 model with TensorFlow Object Detection API, achieving 92% accuracy in static and real-time classification.

Project 3

Earthquake Analysis

Analyzed over 500,000 global earthquakes with magnitudes greater than 4 to explore their correlations with mines and nuclear power plants. Utilized spatial analysis techniques within a 50-mile radius of epicenters to identify key risk factors and trends. Created detailed visualizations to present findings and provided strategic risk mitigation recommendations to five countries. The project employed advanced data processing methods to deliver actionable insights and enhance earthquake risk management.

Project 4

Ad Campaign Success Prediction and ROI Optimization

This project optimized ad campaign performance by predicting success and improving ROI. I built a LightGBM model with 95% accuracy, tuned using GridSearchCV, and selected the top 20 features with RandomForestClassifier. By handling class imbalance with SMOTE and targeting predicted depositors, I increased ROI by 70%, demonstrating the impact of precision-targeted campaigns.

Read My

Medium Articles

Article 1

Understanding Linear regression

In this article, I demystify linear regression, breaking down its fundamental concepts and applications. Through a weather prediction example, I explain simple linear regression, illustrating how it can be used to model relationships between variables and make predictions.

Read More
Article 2

Introduction to API's

In this beginner's guide, I introduce the concept of APIs, explaining how they enable communication between software applications. The article covers key concepts such as API endpoints, methods, and authentication, with practical examples to illustrate their usage. It’s crafted to provide a clear understanding for those new to APIs, making it easier to grasp their importance in modern software development.

Read More
Article 3

Text Preprocessing in NLP: Everything You Need to Know!

In this article I explain essential text preprocessing techniques in NLP with practical code examples. This guide covers everything from tokenization to lemmatization, helping you prepare your text data for machine learning models effectively.

Read More

Get in Touch

Contact Me