The COMSYS Hackathon-4, organized by the COMSYS Educational Trust, Kolkata, and CodeClub JU, was part of the 5th International Conference on Frontiers in Computing (COMSYS-2024) at BITS Pilani, Goa, India, on December 15, 2024. This event benchmarked machine learning and deep learning techniques in data-constrained scenarios, emphasizing robust algorithm development under limited training resources. It featured two tracks: Few-Shot Video Genre Classification and Audio Identification Quest. The first track required classifying 15-second video clips into six sports genres using few-shot learning, while the second one involved classifying audio clips by age group and gender, tackling speech variability. With 523 participants from 109 institutions and 253 teams, the competition, hosted on Kaggle, was evaluated using accuracy (Track 1) and F1 score (Track 2). Finalists presented their approaches after a rigorous verification process. This report details the Hackathon and introduces two publicly available datasets: Track 1– FCS-JU and Track 2– PIS-JU .