• Implementing an evolving IVF dataset
• Fast indexing of 2M vectors for max inner product search
• Limited size clustering
• Asymmetric binary search
• Manual training of IVFPQ
• Mixed sparse-dense clustering
• Separate coarse quantization
• Manual implementation of IVFPQ search
• Sharded GPU dataset with simultaneous queries (pytorch)