Top AI Repos — open-source AI, indexed and scored
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Top AI Repos tracks AI repositories on GitHub and answers two different questions about each one: is it moving right now, and would you bet a product on it.
Curated collection of papers in machine learning systems
| Date | Stars |
|---|---|
| 2026-07-31 | 638 |
| 2026-08-02 | 639 |
| 2026-08-06 | 639 |
Today
— stars today
This week
— stars this week
This month
— stars this month
Momentum
0.0
growth rate 0.00%/day
# Paper List for Machine Learning Systems
[](https://github.com/byungsoo-oh/ml-systems-papers/pulls)
Paper list for broad topics in machine learning systems
> NOTE: Survey papers are annotated with [Survey 🔍] prefix.
## Table of Contents
<!-- TOC -->
- [Paper List for Machine Learning Systems](#paper-list-for-machine-learning-systems)
- [Table of Contents](#table-of-contents)
- [Data Processing](#data-processing)
- [Data pipeline optimization](#data-pipeline-optimization)
- [Caching and distributed storage for ML training](#caching-and-distributed-storage-for-ml-training)
- [LLM data plane](#llm-data-plane)
- [Others](#others)
- [Training System](#training-system)
- [ML job analysis on GPU clusters](#ml-job-analysis-on-gpu-clusters)
- [Resource scheduling](#resource-scheduling)
- [Distributed training](#distributed-training)
- [AutoML](#automl)
- [GNN training system](#gnn-training-system)
- [Inference System](#inference-system)
- [Attention Optimization](#attention-optimization)
- [Mixture of Experts (MoE)](#mixture-of-experts-moe)
- [Communication Optimization \& Network Infrastructure for Distributed ML](#communication-optimization--network-infrastructure-for-distributed-ml)
- [Fault tolerance \& Straggler mitigation](#fault-tolerance--straggler-mitigation)
- [GPU Memory Management \& Optimization](#gpu-memory-management--optimization)
- [GPU Sharing](#gpu-sharing)
- [Compiler](#compiler)
- [GPU Kernel Optimization](#gpu-kernel-optimization)
- [LLM Long Context](#llm-long-context)
- [Model Compression](#model-compression)
- [Federated Learning](#federated-learning)
- [Privacy-Preserving ML](#privacy-preserving-ml)
- [ML APIs \& Application-Side Optimization](#ml-apis--application-side-optimization)
- [ML for Systems](#ml-for-systems)
- [Energy Efficiency](#energy-efficiency)
- [Retrieval-Augmented Generation (RAG)](#retrieval-augmented-generation-rag)
- [Simulation](#simulation)
- [Systems for Agentic AI](#systems-for-agentic-ai)
- [RL Post-Training](#rl-post-training)
- [Multimodal](#multimodal)
- [Hybrid LLMs](#hybrid-llms)
- [Others](#others-1)
- [References](#references)
<!-- /TOC -->
## Data Processing
### Data pipeline optimization
**General**
- [arxiv'25] [Scalable and Performant Data Loading](https://arxiv.org/abs/2504.20067)
- [arxiv'25] [OVERLORD: Ultimate Scaling of DataLoader for Multi-Source Large Foundation Model Training](https://arxiv.org/abs/2504.09844)
- [arxiv'25] [The Streaming Batch Model for Efficient and Fault-Tolerant Heterogeneous Execution](https://arxiv.org/abs/2501.12407)
- [arxiv'25] [In-Network Preprocessing of Recommender Systems on Multi-Tenant SmartNICs](https://arxiv.org/abs/2501.12032)
- [VLDB'25] [cedar: Composable and Optimized Machine Learning Input Data Pipelines](https://arxiv.org/abs/2401.08895)
- [HotInfra'24] [Lotus: Characterize Architecture Level CPU-based Preprocessing in Machine Learning Pipelines](https://kexinrong.github.io/lab/files/lotus-hotinfra24.pdf)
- [arxiv'24] [TensorSocket: Shared Data Loading for Deep Learning Training](https://arxiv.org/abs/2409.18749)
- [arxiv'24] [Efficient Tabular Data Preprocessing of ML Pipelines](https://arxiv.org/abs/2409.14912)
- [MLSys'22] Plumber: Diagnosing and Removing Performance Bottlenecks in Machine Learning Data Pipelines
- [ISCA'22] Understanding Data Storage and Ingestion for Large-Scale Deep Recommendation Model Training
- [SIGMOD'22] Where Is My Training Bottleneck? Hidden Trade-Offs in Deep Learning Preprocessing Pipelines
- [VLDB'21] Analyzing and Mitigating Data Stalls in DNN Training
- [VLDB'21] tf.data: A Machine Learning Data Processing Framework
**Preprocessing stalls**
- [arxiv'24] [PREBA: A Hardware/Software Co-Design for Multi-Instance GPU based AI Inference Servers](https://arxiv.org/abs/2411.19114)
- [ATC'24]Excerpt of 237,225 characters
Read on GitHub
42
Jintao Zhang · @thu-ml, Tsinghua University · China
4
Would you bet a product on this? Bounded 0–100 and slow moving.
matched fp:d82ebad219c038bb, topic:awesome-list