The advancement of autonomous systems, including self-driving and robotics depends on diverse, high-quality datasets. While existing datasets often focus on standard driving scenarios, they frequently lack challenging edge cases, particularly those involving Vulnerable Road Users (VRUs) in complex and dynamic roadside environments. To address this gap, we introduce a novel Small Multimodal Indian Dataset for Learning and Exploration (SMILE) captured in the unique Indian context, showcasing a level of traffic complexity and diversity underrepresented in current benchmarks. We incorporate synchronized data from LiDAR, a stereo camera, and a monocular camera. This resource aims to facilitate the development of more robust autonomous systems. Additionally, we provide a baseline for depth estimation and set a benchmark for future research.