Project Summary: This renewal application aims to support the PI in developing an intelligent ultrafast bioimaging program that could open a new area of investigation and lead to a series of fundamental scientific discoveries. Space and time, two critical physical dimensions, constitute the basis of modern metrology. In bio- imaging, as recognized by the 2014 Nobel Prize in chemistry, there have been breathtaking advances in improving spatial resolution, resulting in an impressive arsenal of nanoscopic tools that can break the diffraction limit of light. Despite the importance, pursuing a high temporal resolution has only recently gained attention. The motivation to develop methods for ultrafast imaging originates from the landscape shift of contemporary biology from morphological explorations and phenotypic probing of organisms to seeking quantitative insights into underlying mechanisms at molecular levels. The transient molecular events occur at a timescale varying from a few nanoseconds that molecules glow in their "fluorescence lifetime" to tens of femtoseconds that molecules take to vibrate. Therefore, ultrafast imaging is essential for observing and characterizing such dynamic events. With support from an ESI MIRA grant, we have established a highly productive research program and developed several ultrafast imaging tools that showed great potential for biological studies. However, several key challenges are still yet to be addressed, such as low light-detection sensitivity at high speed. To solve these problems and further expand the realm of our program, I propose to bring in artificial intelligence and explore a new research direction, intelligent ultrafast bioimaging. Our core idea is to treat an optical imaging system as an encoder-decoder pair, where the imaging system optically encodes a scene on the sensor by physical measurements; an electronic decoder then estimates the object's properties from the raw sensor data. Rather than directly measuring the properties of a biological system, we will acquire encoded measurements through a computational imaging scheme. We will also develop physics-informed machine-learning methods to seamlessly integrate data with physics models in biological systems and codesign the machine-learning algorithms with ultrafast imaging hardware in a holistic manner as an end-to-end optimization problem. We expect that such synergy will lead to a new generation of ultrafast imagers and ultimately reshape many fields in biomedical research.