Hi, my name is Gayatri Khandagale
I am the Data Scientist you need.

I designed a high-speed object detection model using YOLOv8, an advanced variant of the renowned YOLO algorithm, known for its impressive speed (1000x faster than R-CNN) and accuracy. This project focuses on detecting and recognizing American Sign Language (ASL) alphabets from both images and videos, achieving precision and recall rates of 90% and 85% respectively. Trained on the Roboflow ASL dataset, I developed the model for an NGO to facilitate oral exams in the absence of interpreters.

Harnessed the power of OpenCV to normalize retina images from diverse clinics, applied advanced image preprocessing techniques like Gaussian smoothing and circular cropping to standardize the eye shape. By leveraging transfer learning with the ImageNet dataset, I fine-tuned a CNN using EfficientNet architecture in PyTorch, replicating the diagnostic process to detect diabetic retinopathy. With visual cues like abnormal blood vessels and hard exudates in focus, I ensured precision by implementing learning rate schedulers and test-time augmentation to overcome data limitations."

I developed a highly accurate driver drowsiness detection system by combining yawning and non-yawning face images from the YawDD dataset with open and closed eye images from the CEW dataset. Leveraging OpenCV, TensorFlow, and Keras, I designed a deep learning model with convolutional, max pooling, and dense layers. Trained over 50 epochs, the model achieved an impressive 97.06% training accuracy and 97.62% testing accuracy, using 'adam' as the optimizer and 'accuracy' as the performance metric. This innovative project showcases my expertise in computer vision and deep learning.

I developed a word vector generation model using GloVe architecture similar to word2Vec and evaluated the model on the Movie Review dataset. After that, I designed and implemented a TensorFlow graph for a Recurrent Neural Network (RNN) with Long Short-Term Memory (LSTM) units, optimized hyperparameters (LSTM units, optimizer, learning rate/time steps, word vector size) for the best performance, and trained the model on the IMDB dataset. I achieved an accuracy of 96% on the test dataset, showcasing the model's ability to predict sentiment with high accuracy.

This Power BI dashboard provides Plant Co. with actionable insights into gross profit performance, highlighting pain points and growth areas through an interactive, intuitive layout. Starting with raw data in Excel, I transformed it in Power Query and created DAX-driven measures for clear, impactful analysis. SWITCH measures and conditional formatting enhance readability, while thorough testing ensures accuracy, making this dashboard a reliable tool for real-time business insights.

In this project, I developed a dynamic Tableau dashboard analyzing London’s bike rides data. Starting with programmatic data gathering, I used Python’s Pandas for data exploration and transformation, ensuring a clean, insightful dataset. In Tableau, I applied set actions, dynamic parameters, and calculated fields to create an interactive experience that allows users to explore ridership trends, peak times, and popular routes in London. This dashboard offers a powerful, user-driven tool for understanding urban mobility patterns.

Developed a dynamic and interactive Coffee Sales Dashboard in Excel to demonstrate advanced data visualization skills, from foundational data transformation to insightful visual representation. Leveraging Excel’s powerful Pivot Tables and Pivot Charts, I designed an intuitive dashboard that enables easy exploration of sales metrics, trends, and key performance indicators. My hands-on approach highlights a solid grasp of core Excel capabilities, showcasing my ability to turn data into actionable insights and supporting strategic decision-making.

I worked on Microsoft Dynamics 365 applications. Developed .NET Console Apps for high-performance bulk CRUD operations, enhancing business logic with SQL and C# plugins. Also, I led the deployment of Azure Functions, Logic Apps, and Web Jobs/Web Apps to optimize project infrastructure. Additionally, I delivered data insights through reports like Work Order Summary, Service Agreement Compliance, etc. developed in Power BI and Tableau. I even managed on-premise to Azure Cloud migrations using Azure Migrate and Database Migration Services. I implemented CI/CD pipelines via Jenkins, reducing software deployment times and increasing team productivity by 15%. For my contributions, I was awarded Leadership Award for achieving a perfect 3.0/3.0 CSI (Customer Service Index) team rating by implementing Agile methodologies

Gained experience in Databricks to conduct Machine Learning in R & Python, Data engineering, SQL querying on multi-sourced Real estate data. I introduced Sagemaker for Model Training server management on AWS that resulted in cost reduction by 10-12%. I implemented Infrastructure as Code architecture on Terraform, migrated from CloudFormation to use AWS and Azure providers. I also Constructed Github Actions for CI server and monitored CI/CD pipeline.