Hrishikesh Singh

Hello,

I’m Hrishikesh Singh (pronunciation 🔊), a machine learning researcher working in computer vision and deep learning. My research asks how learning systems can perceive, reason over, and act on complex real-world data - especially when that data is noisy, high-dimensional, or scarce.

I’m currently a Research Fellow at IIT Roorkee on a DRDO-funded project, where I develop algorithms for infrared small target detection and multi-frame tracking. This work centers on robust visual perception under low-signal, data-scarce conditions, combining temporal modeling and synthetic data generation to push detection performance where labeled data is limited.

Alongside my research, I co-founded EkLabs, where we build systems that turn unstructured information into structured, actionable knowledge using large language models, knowledge graphs, retrieval, and agentic workflows - extending my interest in perception toward machine reasoning over knowledge.

Earlier, I worked on geospatial perception with LiDAR and 3D mapping, and on reinforcement learning for sequential decision-making. These projects shaped a consistent interest in how learning systems generalize across perception and decision-making.

For my future research, I’m most drawn to multimodal learning and representation learning, and to problems that connect visual perception, reasoning, and decision-making into systems that generalize reliably beyond their training data.