Vanshika Parkar

Curated a high-quality instructional database for research experiments on ASL with video-annotation multimodal data stored in AWS cloud storage, processed with ETL data pipelines in Python, and made the process easier using Airflow jobs. Leveraged AI-augmented analysis and prompt engineering using OpenAI API, Hugging Face LLMs to improve data quality by 15% with NLP tasks on text embeddings and performed data segmentation later for more labeled structure. Generated research standard reports and dashboards in Tableau from experimental results received in Weights and Biases charts, aligning semantic meaning behind the evaluation metrics that made team collaborative discussions easier.