David Tarjan

David Tarjan is a principle research scientist (deep learning) in the Applied Deep Learning Research group at NVIDIA. He has been a performance lead for DLSS SR, FG and RR, including the DLSS Transformer model in DLSS 4. In a prior role he contributed to the Volta architecture.
Avatar photo

Posts by David Tarjan

Three icons, with text LLMs, Optimize, Deploy.
Agentic AI / Generative AI Jan 24, 2025

Dynamic Memory Compression

Despite the success of large language models (LLMs) as general-purpose AI tools, their high demand for computational resources make their deployment challenging... 9 MIN READ