Hi, I’m Sudaksh Kalra 👋 PhD researcher at the University of Amsterdam (UvA) specializing in EdgeAI, dynamic/adaptive neural networks, and hardware-software co-design for AI.
Research Focus: Elastic Neural Networks that could utilize architectural reconfiguration to create a dynamic computation graph for runtime adaptive inference with reduced resource utilization in response to various external changes such as fluctuations in device characterstics, fluctutating requirements, dynamic input quality, and response quality.
Interests: Elastic Language Models, Adaptive Inference, Dynamic Neural Networks, Optimizing neural networks—especially Transformers and LLMs—for reduced computation and energy consumption, as AI agents increasingly integrate into our daily lives.
Currently learning: Efficient computation and hardware-aware DNN inference.
Open to collaboration: Projects aligned with my expertise to deepen knowledge and drive meaningful impact.
Connect with me: LinkedIn