Welcome to the DLER website.

Welcome to the Deep Learning Efficiency Research (DLER) team at NVIDIA Research led by Dr. Pavlo Molchanov.

Founded in 2023, we are a vibrant and dynamic research lab. We operate under the umbrella of the Learning and Perception Research (LPR) group, led by Dr. Jan Kautz.

Our mission is to drive advancements in the efficiency of deep learning technologies. In the hardware layer, we focus on reducing memory footprint, minimizing inference latency, and lowering energy consumption of models. In the software layer, we prioritize the development of efficient small models and the design of reliable agentic systems.

Currently, we study:

• novel and efficient model architectures;
• post-training model optimization with techniques such as compression, sparsity, quantization, and neural architecture search;
• powerful, universal representations of images and videos;
• efficiency and reliability of agentic systems;
• efficiency and reliability of applications of language models and multi-modal models in intelligent assistants, robotics, and autonomous driving.

We solicit applications from exceptional deep learning researchers. Explore the opportunities at the NVIDIA Careers website or reach out directly to our team members via email.