About the job
Gaussian Splatting (GS) is a 3D/4D scene reconstruction technique that enables photorealistic novel-view synthesis with low rendering complexity, making it attractive for deployment on consumer devices such as TVs, streaming sticks, phones, and laptops. Realizing this vision requires addressing several open technical challenges, such as a significant reduction in model training/encoding time and more efficient compression. As part of the Video Algorithms team during this 24-week Fall internship, you will help us investigate the potential of GS as a future streaming format and explore possible improvements, with a focus on building towards a practical system.
Responsibilities
Explore GS model compression strategies using open datasets
Contribute to early thinking on additional dataset needs for representative scenes.
Characterize trade-offs among GS model size, training time, and rendered quality, and quantify the gap relative to streaming-rate targets
Identify and experiment with strategies to reduce training/encoding time and/or to improve GS compression efficiency
Design and implement a proof-of-concept (PoC) that showcases GS-based rendering on content of interest
Qualifications
Minimum
Currently pursuing a PhD in a technical field such as Computer Science, Engineering, Math, or Statistics, with an expected graduation date in June 2027 or later.
Thrives working in complex, dynamic, and fast-moving environments.
Strong software development skills and feels comfortable with software engineering best practices (e.g., version control, testing, code review, etc.).
Successful track record in research of 3D/4D scene reconstruction, novel-view synthesis, Gaussian Splatting or NeRF, differentiable rendering, neural graphics, or 3D computer vision.
Solid understanding of machine learning and deep learning concepts, with hands-on experience training and evaluating ML models.
Able to program fluently in Python
Preferred
Familiarity with real-time rendering and GPU programming (CUDA, WebGL, graphics pipelines).
Background in video compression, streaming systems, or codec standards such as HEVC and AV1.
Involvement in open-source multimedia or graphics projects.
Experience with large-scale distributed systems and cloud computing.