Haricharan Aragonda
M.Tech Research Graduate
Biography
Haricharan A is an applied machine learning researcher who has worked in various areas such as new born hearing screening, ad-tech, e-commerce and video telematics at scale with a strong focus on computer vision, video understanding, and representation learning. His work sits at the intersection of theory and practice, with an emphasis on building models and pipelines from first principles.
Over the years, his research and applied work have spanned deep learning for visual perception, temporal modeling of video data, multimodal learning, and optimization techniques for real-world constraints such as limited data, class imbalance, and deployment efficiency. He has hands-on experience implementing custom architectures, loss functions, and training strategies, and is comfortable moving from mathematical intuition to robust, production-ready systems.
Haricharan's broader research philosophy is grounded in simplicity, interpretability, and careful problem formulation. He is especially drawn to problems where classical ideas such as signal processing, probability, and geometry meet modern deep learning approaches. Beyond model development, he has a keen interest in mentoring, technical communication, and breaking down complex concepts for diverse audiences.
He continues to explore challenging problems in vision and multimodal learning, with the goal of building systems that are not only accurate, but reliable, efficient, and well understood.