A modern replacement for Redis and Memcached

Repo for ECHO: Efficient KV Cache Offloading with Lossless Prefetching for Serving Native Sparse Attention LLMs (OSDI'26)

DeepInfer: High-performance, large-scale inference framework for DeepSeek models

Credo: Declarative Control of LLM Pipelines via Beliefs and Policies (VLDB 2026 Demo)

An In-Depth Study of Filter-Agnostic Vector Search on a PostgreSQL Database System (SIGMOD 2026)

Lightweight coding agent that runs in your terminal

AI-Driven Scientific, Algorithmic, and Systems Discovery

A framework for efficient model inference with omni-modality models

Benchmarking Open-Ended Inference Optimization by AI Agents

DeepSeek 4 Flash and PRO local inference engine for Metal, CUDA and ROCm

Aker: Density-Aware Approximate Caching for Vector Search

Conditional Memory via Scalable Lookup: A New Axis of Sparsity for Large Language Models

AI agents running research on single-GPU nanochat training automatically

Memory Sparse Attention - A scalable, end-to-end trainable latent-memory framework for 100M-token contexts.

High Performance Serving Runtime For Interactive Super Intelligence System

The agent that grows with you

Ongoing research training transformer models at scale

AgentFlow: In-the-Flow Agentic System Optimization

I seek to draw from the philosophies embedded in AI research to inspire deeper reflection on life — exploring how the principles that guide intelligent systems can illuminate the way we live, think…

Design hardware-friendly model architectures and migrate existing LLMs with minimal performance loss

[ICLR 2026] RPG: A Repository Planning Graph for Unified and Scalable Codebase Generation

A compact implementation of SGLang, designed to demystify the complexities of modern LLM serving systems.

Bespoke OLAP: Synthesizing Workload-Specific One-size-fits-one Database Engines

GenDB, an LLM-Powered Generative Query Engine Built for the Future

Evolve your language agent with Agentic Context Engineering (ACE)

Dynamic Cheatsheet: Test-Time Learning with Adaptive Memory

[NeurIPS 2023] Tree of Thoughts: Deliberate Problem Solving with Large Language Models