In today's episode of "High Impact Thesis," we sit down with Alka Luqman, a final-year PhD student at NTU's College of Computing and Data Science. Before starting her PhD, Alka spent seven years in an industrial company, where she worked on machine learning projects and natural language processing in the pre-LLM era. Alka shares her motivation for leaving the corporate world, to create new technology rather than just use it. Her current research focuses on peer-to-peer federated learning, a form of distributed machine learning that allows data to remain on individual devices, improving privacy and security. We discuss the challenges of implementing this technology in the real world, including data distribution shifts and communication efficiency, which is a core focus of her research.