GPU-Based Streaming Incremental SVD Implementation

I am looking for an expert in GPU computing and Dask who can implement Incremental Singular Value Decomposition (SVD) on multiple GPUs. The primary goal of this project is to handle larger datasets, specifically image data, using distributed streaming SVD. Key Requirements: - Implement Dask_GPU for improved parallel processing - Optimize the system for scalability across multiple GPUs - Ensure the system can handle large image datasets efficiently - Implement efficient memory management for large datasets on GPUs. - Develop robust error handling mechanisms for distributed computations. The ideal candidate for this project would have extensive experience in distributed computing, particularly with Dask and GPU. A deep understanding of SVD and its incremental variant is crucial. Experience working with image data is a plus. Note : Max Budget - 5K Visit [login to view URL] for more clarity on concept. Already done work : Distributed SVD computations is already done in Dask. What required is Distributed streaming SVD . Check the files I have shared , in the jupyter notebook file do check the markdown section "Start from here" to "Ended here". You can run the code on Kaggle.