Gursimar Wadhawan

Professors : - Dr. Ram Babu Karumudi

• Enhanced image resolution by 400% using state-of-the-art deep learning techniques, such as super-resolution autoencoders with 92% accuracy and SRCNN for high-precision upscaling in quantum simulator images.
• Applied advanced pattern recognition algorithms to identify intricate patterns, including honeycomb and grid structures, in Near Field and Far Field experimental images, significantly contributing to the advancement of quantum simulator performance.
• Integrated computer vision methods, including contour detection, Hough circle transformation, and edge detection, to improve the accuracy of deep learning models, resulting in a 35% increase in pattern detection precision for complex quantum imaging tasks.

- Spearheaded a BERTopic-driven NLP solution by integrating sentence transformers and HDBSCAN clustering on a dataset of approximately 50,000 tweets, achieving a model coherence score of 0.73 and successfully removing 100% of outliers from the labeled dataset.
- Designed and developed data collection frameworks for various social media platforms using APIs such as YouTube Data V3, enabling the analysis of data to combat misinformation on topics including propaganda, climate change, diet myths, and public health.
- Developed an XGBoost-powered time series forecasting model on a comprehensive dataset with over 500,000 data points per household, accurately predicting indoor temperatures and heatwaves with a 93% accuracy rate for numerous North American households, and assessing their vulnerability and risk scores.
- Implemented an emotion prediction model using LSTM on user data from health apps and linked wearables (Fitbit, Apple Watch, Garmin, etc.), applying feature engineering and data mining to extract actionable dimensions.
- Demonstrated expertise in the Azure and Databricks platforms for cloud-based data processing and deployment, driving efficient data analysis and operational efficiency.