Jigyasa Singhal

Upgraded and deployed a resolution algorithm, boosting automation by 3%, saving $1.5M in BPO costs. Enhanced PySpark data preprocessing, achieving 8x speed-up and reducing costs by 25%. Served as Scrum Master for Data Science support team, overseeing agile processes and maintaining JIRA boards. Reduced ticket resolution time by 20% through efficient communication and collaboration with clients.

Collaborated with Clinical product teams to process 500k documents per day for Classification via Tensorflow Conv-Attention model. Orchestrated end-to-end ML model retraining pipelines on Kubernetes using Docker images corresponding to data extraction, preprocessing, training, evaluation, and inference components. Used AWS to deliver ML solutions at scale, managed multiple ML models in production covering all aspects of model life cycle management. Delivered project presentations to non-technical stakeholders, simplifying complex technical concepts such as model architectures and deployment strategies.

Created a human-computer interface BCI model that enables direct communication between humans and computers through brain measurement analysis. Transformed the problem into binary classification and implemented a deep convolutional neural network for P300 speller, achieving an impressive accuracy of 95%, surpassing previous models at 86.8%.