Parampuneet Kaur Thind

Information

Bio

I am Parampuneet Kaur Thind (Param), a professional in the field of hardware-aware AI and deep learning optimization with a solid foundation in embedded systems and edge computing. I am currently a National PhD student at Sapienza Università di Roma (DNOT 40th cycle). With my expertise, I have contributed to Little Place Labs Inc. for 3 impactful years, focusing on Neural Architecture Search (NAS), model compression, and AI deployment for space applications. My work bridges the gap between AI advancements and hardware constraints, enabling efficient onboard processing for Earth Observation satellites.

Current Role

Currently, I serve as a Data Scientist at Little Place Labs Inc., where I focus on developing and optimizing AI models for edge computing and satellite-based applications. At Little Place Labs, I am dedicated to advancing hardware-aware AI, leveraging Neural Architecture Search (NAS) and model compression techniques to enhance onboard processing capabilities for Earth Observation satellites.

Areas of Expertise

My expertise lies in several key areas of hardware-aware AI and deep learning optimization, including Neural Architecture Search (NAS), model compression techniques, and AI deployment on embedded systems. These specializations enable me to design and optimize deep learning models that efficiently operate within the constraints of edge and spaceborne hardware, ensuring high-performance AI for onboard satellite processing. This expertise can significantly benefit ESA Φ-lab CIN by advancing autonomous AI-driven Earth Observation capabilities, improving onboard inference efficiency, and enabling real-time AI applications in space.

Vision for the Future

My vision for the future is to redefine AI deployment in space by making onboard intelligence more autonomous, efficient, and adaptive to real-world constraints. As AI continues to play a crucial role in Earth Observation (EO) and remote sensing, I aim to push the boundaries of hardware-aware AI, ensuring that deep learning models are not only accurate but also optimized for real-time inference on spaceborne and edge devices.

By leveraging Neural Architecture Search (NAS), model compression, and hardware-efficient AI, my work contributes to reducing data transmission bottlenecks, enhancing real-time decision-making in space, and enabling fully autonomous satellite AI systems. This will revolutionize EO applications, allowing satellites to process and analyze data onboard, detect anomalies instantly, and trigger intelligent responses without relying solely on ground stations.

Looking ahead, I envision AI models that dynamically adapt to changing mission needs, optimizing themselves in space based on available resources and environmental conditions. This will pave the way for the next generation of self-learning, resilient AI systems for deep space exploration, planetary missions, and autonomous EO platforms, ultimately enhancing global environmental monitoring, disaster response, and climate research.