| 1 | # Committers |
| 2 | |
| 3 | This document lists the current committers of the vLLM project and the core areas they maintain. |
| 4 | Committers have write access to the vLLM repository and are responsible for reviewing and merging PRs. |
| 5 | You can also refer to the [CODEOWNERS](https://github.com/vllm-project/vllm/blob/main/.github/CODEOWNERS) file for concrete file-level ownership and reviewers. Both these documents and the CODEOWNERS file are living documents and they complement each other. |
| 6 | |
| 7 | ## Active Committers |
| 8 | |
| 9 | We try to summarize each committer's role in vLLM in a few words. In general, vLLM committers cover a wide range of areas and help each other in the maintenance process. |
| 10 | Please refer to the later section about Area Owners for exact component ownership details. |
| 11 | Sorted alphabetically by GitHub handle: |
| 12 | |
| 13 | - [@aarnphm](https://github.com/aarnphm): Structured output |
| 14 | - [@AndreasKaratzas](https://github.com/AndreasKaratzas): ROCm / AMD GPU integration, CI |
| 15 | - [@ApostaC](https://github.com/ApostaC): Connectors, offloading |
| 16 | - [@bbrowning](https://github.com/bbrowning): Tool use and reasoning parser |
| 17 | - [@benchislett](https://github.com/benchislett): Engine core and spec decode |
| 18 | - [@bigPYJ1151](https://github.com/bigPYJ1151): Intel CPU/XPU integration |
| 19 | - [@BoyuanFeng](https://github.com/BoyuanFeng): Compilation, CUDAGraph |
| 20 | - [@BugenZhao](https://github.com/BugenZhao): Rust frontend |
| 21 | - [@chaunceyjiang](https://github.com/chaunceyjiang): Tool use and reasoning parser |
| 22 | - [@DarkLight1337](https://github.com/DarkLight1337): Multimodality, API server |
| 23 | - [@dllehr-amd](https://github.com/dllehr-amd): AMD integration |
| 24 | - [@esmeetu](https://github.com/esmeetu): developer marketing, community |
| 25 | - [@gau-nernst](https://github.com/gau-nernst): Kernels and performance |
| 26 | - [@gshtras](https://github.com/gshtras): ROCm / AMD GPU integration |
| 27 | - [@Harry-Chen](https://github.com/Harry-Chen): Build, CI, CUDA architecture |
| 28 | - [@heheda12345](https://github.com/heheda12345): Hybrid memory allocator |
| 29 | - [@hmellor](https://github.com/hmellor): Hugging Face integration, documentation |
| 30 | - [@hongxiayang](https://github.com/hongxiayang): AMD GPU / ROCm integration |
| 31 | - [@houseroad](https://github.com/houseroad): Engine core and Llama models |
| 32 | - [@Isotr0py](https://github.com/Isotr0py): Multimodality, new model support |
| 33 | - [@ivanium](https://github.com/ivanium): KV Connector and offload |
| 34 | - [@jeejeelee](https://github.com/jeejeelee): LoRA, new model support |
| 35 | - [@jikunshang](https://github.com/jikunshang): Intel CPU/XPU integration |
| 36 | - [@jperezdealgaba](https://github.com/jperezdealgaba): Security |
| 37 | - [@khluu](https://github.com/khluu): CI infrastructure |
| 38 | - [@LucasWilkinson](https://github.com/LucasWilkinson): Kernels and performance |
| 39 | - [@markmc](https://github.com/markmc): Observability |
| 40 | - [@MatthewBonanni](https://github.com/MatthewBonanni): Kernels and performance |
| 41 | - [@mgoin](https://github.com/mgoin): Quantization and performance |
| 42 | - [@NickLucche](https://github.com/NickLucche): KV connector |
| 43 | - [@njhill](https://github.com/njhill): Distributed, API server, engine core |
| 44 | - [@noooop](https://github.com/noooop): Pooling models |
| 45 | - [@orozery](https://github.com/orozery): KV Connector and offload |
| 46 | - [@pavanimajety](https://github.com/pavanimajety): NVIDIA GPU integration |
| 47 | - [@ProExpertProg](https://github.com/ProExpertProg): Compilation, startup UX |
| 48 | - [@robertgshaw2-redhat](https://github.com/robertgshaw2-redhat): Core, distributed, disagg |
| 49 | - [@sfeng33](https://github.com/sfeng33): Tool use and reasoning parser |
| 50 | - [@shen-shanshan](https://github.com/shen-shanshan): AMD GPU / ROCm integration, multimodality, ViT CUDA graph |
| 51 | - [@simon-mo](https://github.com/simon-mo): Project lead, API entrypoints, community |
| 52 | - [@tdoublep](https://github.com/tdoublep): State space models |
| 53 | - [@tjtanaa](https://github.com/tjtanaa): AMD GPU integration |
| 54 | - [@tlrmchlsmth](https://github.com/tlrmchlsmth): Kernels and performance, distributed, disagg |
| 55 | - [@tomeras91](https://github.com/tomeras91): MoE, Mamba, Ray |
| 56 | - [@vadiklyutiy](https://github.com/vadiklyutiy): CI, performance, kernels |
| 57 | - [@WoosukKwon](https://github.com/WoosukKwon): Project lead, engine core |
| 58 | - [@wzhao18](https://github.com/wzhao18): Kernels and performance, KV offloading |
| 59 | - [@xinli-sw](https://github.com/xinli-sw): Quantization and model support |
| 60 | - [@xuechendi](https://github.com/xuechendi): Intel CPU/XPU integration, KV connector |
| 61 | - [@yaochengji](https://github.com/yaochengji): TPU integration |
| 62 | - [@yewentao256](https://github.com/yewentao256): Kernels and performance |
| 63 | - [@Yikun](https://github.com/Yikun): Pluggable hardware interface |
| 64 | - [@youkaichao](https://github.com/youkaichao): Project lead, distributed, compile, community |
| 65 | - [@ywang96](https://github.com/ywang96): Multimodality, benchmarks |
| 66 | - [@zhuohan123](https://github.com/zhuohan123): Project lead, RL integration, numerics |
| 67 | - [@ZJY0516](https://github.com/ZJY0516): State space models, spec decode, new model support |
| 68 | - [@zou3519](https://github.com/zou3519): Compilation |
| 69 | - [@zyongye](https://github.com/zyongye): MoE kernels and quantization |
| 70 | |
| 71 | ### Emeritus Committers |
| 72 | |
| 73 | Committers who have contributed to vLLM significantly in the past (thank you!) but no longer active: |
| 74 | |
| 75 | - [@22quinn](https://github.com/22quinn): RL API |
| 76 | - [@alexm-redhat](https://github.com/alexm-redhat): Performance |
| 77 | - [@andoorve](https://github.com/andoorve): Pipeline parallelism |
| 78 | - [@cadedaniel](https://github.com/cadedaniel): Speculative decoding |
| 79 | - [@comaniac](https://github.com/comaniac): KV cache management, pipeline parallelism |
| 80 | - [@KuntaiDu](https://github.com/KuntaiDu): KV Connector |
| 81 | - [@LiuXiaoxuanPKU](https://github.com/LiuXiaoxuanPKU): Speculative decoding |
| 82 | - [@luccafong](https://github.com/luccafong): Llama models, speculative decoding, distributed |
| 83 | - [@patrickvonplaten](https://github.com/patrickvonplaten): Mistral models, new model support |
| 84 | - [@pcmoritz](https://github.com/pcmoritz): MoE |
| 85 | - [@rkooo567](https://github.com/rkooo567): Chunked prefill |
| 86 | - [@ruisearch42](https://github.com/ruisearch42): Pipeline parallelism, Ray Support |
| 87 | - [@russellb](https://github.com/russellb): Structured output, engine core, security |
| 88 | - [@sighingnow](https://github.com/sighingnow): Qwen models, new model support |
| 89 | - [@sroy745](https://github.com/sroy745): Speculative decoding |
| 90 | - [@Yard1](https://github.com/Yard1): kernels and performance |
| 91 | - [@yeqcharlotte](https://github.com/yeqcharlotte): Benchmark, Llama models |
| 92 | - [@zhisbug](https://github.com/zhisbug): Arctic models, distributed |
| 93 | |
| 94 | ## Area Owners |
| 95 | |
| 96 | This section breaks down the active committers by vLLM components and lists the area owners. |
| 97 | If you have PRs touching the area, please feel free to ping the area owner for review. |
| 98 | |
| 99 | ### Engine Core |
| 100 | |
| 101 | - Scheduler: the core vLLM engine loop scheduling requests to next batch |
| 102 | - @WoosukKwon, @robertgshaw2-redhat, @njhill, @heheda12345 |
| 103 | - KV Cache Manager: memory management layer within scheduler maintaining KV cache logical block data |
| 104 | - @heheda12345, @WoosukKwon |
| 105 | - AsyncLLM: the zmq based protocol hosting engine core and making it accessible for entrypoints |
| 106 | - @robertgshaw2-redhat, @njhill |
| 107 | - ModelRunner, Executor, Worker: the abstractions for engine wrapping model implementation |
| 108 | - @WoosukKwon, @tlrmchlsmth, @heheda12345, @LucasWilkinson, @ProExpertProg, @MatthewBonanni |
| 109 | - KV Connector: Connector interface and implementation for KV cache offload and transfer |
| 110 | - @robertgshaw2-redhat, @njhill, @NickLucche, @ApostaC, @ivanium, @orozery |
| 111 | - Distributed, Parallelism, Process Management: Process launchers managing each worker, and assign them to the right DP/TP/PP/EP ranks |
| 112 | - @youkaichao, @njhill, @WoosukKwon |
| 113 | - Collectives: the usage of nccl and other communication libraries/kernels |
| 114 | - @tlrmchlsmth, @youkaichao |
| 115 | - Multimodality engine and memory management: core scheduling and memory management concerning vision, audio, and video inputs. |
| 116 | - @ywang96, @DarkLight1337 |
| 117 | |
| 118 | ### Model Implementations |
| 119 | |
| 120 | - Model Interface: The `nn.Module` interface and implementation for various models |
| 121 | - @zhuohan123, @mgoin, @simon-mo, @houseroad, @ywang96 (multimodality), @jeejeelee (lora) |
| 122 | - Logits Processors / Sampler: The provided sampler class and pluggable logits processors |
| 123 | - @njhill, @houseroad |
| 124 | - Custom Layers: Utility layers in vLLM such as rotary embedding and rms norms |
| 125 | - @ProExpertProg |
| 126 | - Attention: Attention interface for paged attention |
| 127 | - @WoosukKwon, @LucasWilkinson, @heheda12345, @MatthewBonanni |
| 128 | - FusedMoE: FusedMoE kernel, Modular kernel framework, EPLB |
| 129 | - @tlrmchlsmth |
| 130 | - Quantization: Various quantization config, weight loading, and kernel. |
| 131 | - @mgoin, @Isotr0py, @yewentao256, @zyongye, @xinli-sw |
| 132 | - Custom quantized GEMM kernels (cutlass_scaled_mm, marlin, machete) |
| 133 | - @tlrmchlsmth, @LucasWilkinson |
| 134 | - Multi-modal Input Processing: Components that load and process image/video/audio data into feature tensors |
| 135 | - @DarkLight1337, @ywang96, @Isotr0py |
| 136 | - torch compile: The torch.compile integration in vLLM, custom passes & transformations |
| 137 | - @ProExpertProg, @zou3519, @youkaichao, @BoyuanFeng |
| 138 | - State space models: The state space models implementation in vLLM |
| 139 | - @tdoublep, @tlrmchlsmth, @ZJY0516, @tomeras91 |
| 140 | - Reasoning and tool calling parsers |
| 141 | - @chaunceyjiang, @aarnphm, @sfeng33, @bbrowning |
| 142 | |
| 143 | ### Entrypoints |
| 144 | |
| 145 | - LLM Class: The LLM class for offline inference |
| 146 | - @DarkLight1337 |
| 147 | - API Server: The OpenAI-compatible API server |
| 148 | - @DarkLight1337, @njhill, @aarnphm, @simon-mo, @heheda12345 (Responses API) |
| 149 | - Rust Frontend: The experimental API server in Rust |
| 150 | - @BugenZhao, @njhill |
| 151 | - Batch Runner: The OpenAI-compatible batch runner |
| 152 | - @simon-mo |
| 153 | |
| 154 | ### Features |
| 155 | |
| 156 | - Spec Decode: Covers model definition, attention, sampler, and scheduler related to n-grams, EAGLE, and MTP. |
| 157 | - @WoosukKwon, @benchislett, @MatthewBonanni |
| 158 | - Structured Output: The structured output implementation |
| 159 | - @aarnphm |
| 160 | - RL: The RL related features such as collective rpc, sleep mode, etc. |
| 161 | - @youkaichao, @zhuohan123 |
| 162 | - LoRA: @jeejeelee |
| 163 | - Observability: Metrics and Logging |
| 164 | - @markmc, @robertgshaw2-redhat, @simon-mo |
| 165 | |
| 166 | ### Code Base |
| 167 | |
| 168 | - Config: Configuration registration and parsing |
| 169 | - @hmellor |
| 170 | - Documentation: @hmellor, @DarkLight1337, @simon-mo |
| 171 | - Benchmarks: @ywang96, @simon-mo |
| 172 | - CI, Build, Release Process: @khluu, @njhill, @simon-mo, @Harry-Chen, @vadiklyutiy |
| 173 | - Security: @jperezdealgaba |
| 174 | |
| 175 | ### External Kernels Integration |
| 176 | |
| 177 | - FlashAttention: @LucasWilkinson, @MatthewBonanni |
| 178 | - FlashInfer: @LucasWilkinson, @mgoin, @WoosukKwon, @MatthewBonanni, @wzhao18 |
| 179 | - Blackwell Kernels: @mgoin, @yewentao256 |
| 180 | - DeepEP/DeepGEMM: @mgoin, @yewentao256 |
| 181 | - FlashKDA: @gau-nernst |
| 182 | |
| 183 | ### Integrations |
| 184 | |
| 185 | - Hugging Face: @hmellor, @Isotr0py |
| 186 | - Ray: (unowned) |
| 187 | - NIXL: @robertgshaw2-redhat, @NickLucche |
| 188 | |
| 189 | ### Hardware |
| 190 | |
| 191 | - Plugin Interface: @youkaichao, @Yikun |
| 192 | - NVIDIA GPU: @pavanimajety, @wzhao18 |
| 193 | - AMD GPU: @gshtras, @tjtanaa, @hongxiayang, @shen-shanshan, @AndreasKaratzas |
| 194 | - Intel CPU/GPU: @jikunshang, @bigPYJ1151, @xuechendi |
| 195 | - Google TPU: @yaochengji |
| 196 | |
| 197 | ### Ecosystem Projects |
| 198 | |
| 199 | - Ascend NPU: [@wangxiyuan](https://github.com/wangxiyuan) and [see more details](https://vllm-ascend.readthedocs.io/en/latest/community/contributors.html#maintainers) |
| 200 | - Intel Gaudi HPU [@xuechendi](https://github.com/xuechendi) and [@iboiko-habana](https://github.com/iboiko-habana) |
| 201 | - Semantic Router: [@xunzhuo](https://github.com/xunzhuo), [@rootfs](https://github.com/rootfs) and [see more details](https://vllm-semantic-router.com/community/team) |