Dr Vallurupalli Vamsi
Dr Vallurupalli Vamsi
Marie Sklodowska-Curie Fellow
University of Galway
Profile
Dr. Vallurupalli Vamsi earned his PhD in Management with a specialization in Management Information Systems (MIS) from the Indian Institute of Management Calcutta, India, in 2019. His doctoral research examined the factors influencing the impact of online reviews, employing natural language processing (NLP) techniques. Broadly, his research focuses on the application of machine learning methods in business contexts, with a particular emphasis on the intersection of data-driven technologies and business processes. His work has been published in leading academic journals, including Decision Support Systems, Communications of the Association for Information Systems, and Electronic Markets.
Following his PhD, Dr. Vamsi worked in the industry for a few years in reputed organizations, holding both technical and managerial roles. Most notably, he served as a Senior Manager (Analytics) at a fintech company that offered personal loans entirely through a mobile platform, where he led the data science team and developed credit risk models, fraud detection systems, and automation pipelines. He also contributed to anti-money laundering and customer analytics initiatives in other roles. After his industry experience, he spent four years as an Assistant Professor at Shiv Nadar Institution of Eminence, India, where he taught courses related to analytics, including Contemporary Business Analytics, Introduction to Artificial Intelligence, Quantitative Methods in Business, and Business Intelligence using Tableau.
He possesses strong technical expertise in machine learning, deep learning, and NLP, and is proficient in Python, SQL, Tableau, Hadoop, and Spark. Additionally, he has a solid grounding in econometrics and working knowledge of the case study research method.
Contact Information
LinkedIn: http://linkedin.com/in/dr-vamsi-vallurupalli-a3544760
ORCID ID: 0009-0002-0829-0463
ROSETTA Project
Fairness in AutoML systems: Identifying (temporal) seams and mitigating them through a human-centered design
Automated Machine Learning (AutoML) systems are seen as a solution to the shortage of ML professionals. With rising calls for responsible ML, AutoML systems are now embedding fairness, and are referred to as Fairness-Aware AutoML (FA-AutoML). While promising, a closer investigation of the current FA-AutoML systems reveals an important concern. Efforts to mitigate fairness issues have mostly been incorporated in the model building stage. Yet, fairness issues can arise at any stage of the development process. Mitigating these issues often requires human input and judgment, creating a paradox: building fair models needs human agency, but AutoML aims to minimize human involvement. There is little work on addressing this paradox, which is a major gap in the literature.
To address this gap, I will first identify points in the ML development life cycle where automated actions may lead to fairness issues, referred to as “seams”. Once these seams are identified, the next step is to develop an approach to mitigate fairness concerns at these points.
In this research I propose that a human-centered design incorporating well-placed interruptions in the ML life cycle could help mitigate the fairness issues. This research will identify the appropriate temporal points and determine suitable characteristics for such interruptions. Additionally, an FA-AutoML system designed to integrate these interruptions will be developed and evaluated. As the process includes the design, development, and testing of an IT artifact to address an organizational problem, the study will follow the design science research method (DSRM).
ROSETTA Supervisor and Host Institution
Supervisor: Dr Anastasia Griva
Host: University of Galway