Transparency is widely recognized to be one of the basic principles of good governance of artificial intelligence (AI). This paper discusses the why and how of transparency obligations, as articulated in the AI governance discussions in India and in select international principles. It argues that the need for transparency permeates through the lifecycle of an AI project and identifies the policy layer, the technical layer and the operational layer as the key sites for fostering the transparency in any AI project. It then studies India’s Digi Yatra project, a system for biometric identity verification at airports, to examine how the project fares on transparency across these three identified layers. The paper points to certain gaps in the discharge of transparency obligations in connection with the Digi Yatra project and its lessons for AI transparency, particularly in the context of large public-facing projects.