Virtual Malloc Logovirtual malloc

Research that moves from theory to production

Our research teams work at the edges of systems, machine learning, and scientific computing — publishing openly and shipping the results into real-world deployments.

40+

Peer-reviewed papers

12

Open-source projects

7

Active research areas

Where we focus

Cross-disciplinary teams pursuing open problems with measurable, deployable outcomes.

High-Performance Systems

Memory allocators, lock-free data structures, and runtime schedulers engineered for predictable performance at scale.

Machine Learning

Efficient training and inference, model compression, and retrieval-augmented architectures for production workloads.

Scientific Computing

Abstraction layers and numerical methods that make large-scale simulation portable across heterogeneous hardware.

Distributed Systems

Consensus, replication, and fault-tolerance primitives for low-latency services spanning regions and clouds.

Security & Cryptography

Applied cryptography, confidential computing, and formal verification of safety-critical software components.

Compilers & Languages

Domain-specific languages and optimizing compilers that close the gap between expressive code and bare-metal speed.

Selected publications

A sample of recent work from our research teams, published at leading venues.

  • OSDI 2026

    Virtualmalloc: A Tier-Aware Allocator for Predictable Tail Latency

  • NeurIPS 2025

    Sparse Retrieval Without the Index: Streaming Memory for Long-Context Models

  • SC 2025

    Portable Abstraction Layers for High-Performance Scientific Simulation

  • SOSP 2024

    Deterministic Scheduling for Multi-Tenant GPU Clusters