SALT Systems Lab

Scalable Algorithms & Systems · Northeastern University

Fast, memory-efficient data systems for modern hardware.

We design data structures and systems that scale with today's hardware: filters and hash tables with provable guarantees, GPU memory allocators and data structures, and GPU-native vector search and serving for AI workloads. All of our projects are open source.

Explore projects github.com/saltsystemslab

Research areas

Our projects fall into four areas that share the same goal: high throughput with very little memory.

Filters

Compact approximate-membership data structures: counting, vectorized, adaptive, and range filters with provable guarantees.

CQFVQFBreadcrumb FilterAdaptiveQFSkiQFAeris Filter

Hash Tables

Fast, concurrent hash tables that stay stable and efficient at high load factors.

IcebergHTZombie Hashing

GPU Data Management

Building blocks for GPU applications: memory management, hash tables, and filters designed for thousands of concurrent threads.

GPU FilterswarpSpeedGallatinAndes

AI Systems

GPU-native infrastructure for vector search, retrieval, and LLM serving.

JasperInferScaleCascade

People

SALT Systems Lab is part of the Khoury College of Computer Sciences at Northeastern University. Interested in working with us? Get in touch through the lab director's page.

Alumni