Skip to content
View luqidaxia's full-sized avatar
  • 力数智能科技(江苏)有限公司

Block or report luqidaxia

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
luqidaxia/README.md

陆琪 · Lu Qi

力数智能科技(江苏)有限公司 · 总经理 General Manager · Lexdata Intelligent Technology (Jiangsu) Co., Ltd.


中文

现任 力数智能科技(江苏)有限公司 总经理。

关注的技术方向:

  • 大模型推理服务化 —— vLLM 生产部署、OpenAI 兼容接口、长上下文与并发调度
  • GPU 集群与互联 —— PCIe / NUMA 拓扑、NCCL 卡间通信、多卡并行策略
  • 推理性能工程 —— 投机解码、KV Cache 显存与并发的权衡、CUDA 图与批调度
  • 新硬件适配 —— Blackwell(SM120)等新架构上的算子分发与补丁维护

English

General Manager at Lexdata Intelligent Technology (Jiangsu) Co., Ltd.

Focus areas:

  • LLM inference serving — production vLLM deployments, OpenAI-compatible APIs, long-context and concurrency scheduling
  • GPU clusters & interconnect — PCIe / NUMA topology, NCCL inter-GPU communication, multi-GPU parallel strategies
  • Inference performance engineering — speculative decoding, KV-cache memory vs. concurrency trade-offs, CUDA graphs and batch scheduling
  • New-architecture enablement — kernel dispatch and patch maintenance on new silicon such as Blackwell (SM120)

代表项目 · Featured

项目 说明
deepseekv41_pro6000d_tp8 单机 8× RTX 6000D 部署 DeepSeek-V4.1-Flash(TP=8):SM120 补丁链、DSpark 投机解码、FP8 KV Cache;含完整的 PCIe 拓扑与 NCCL P2P 实测(同 switch / 跨 switch / 跨 NUMA 三类带宽、NCCL_P2P_LEVEL 档位对照)

力数智能科技(江苏)有限公司 · Lexdata Intelligent Technology (Jiangsu) Co., Ltd.

Popular repositories Loading

  1. dgxspark_comfyui_minimax_h3 dgxspark_comfyui_minimax_h3 Public

    Python 17 5

  2. deepseekv41_pro6000d_tp8 deepseekv41_pro6000d_tp8 Public

    Single-node 8x RTX 6000D deployment of DeepSeek-V4.1-Flash (TP=8): SM120 patches, DSpark speculative decoding, FP8 KV, and measured PCIe topology / NCCL P2P analysis. 力数智能科技(江苏)有限公司 / lexdata

    Python 3

  3. dgxspark_deepseekv4flash0731 dgxspark_deepseekv4flash0731 Public

    Shell 2

  4. luqidaxia luqidaxia Public

    GitHub profile README — 陆琪 (Lu Qi)

  5. dgxspark_deepseekv4.1_4node dgxspark_deepseekv4.1_4node Public

    Shell

  6. qwen3.8-flash-next-nvfp4-tp2 qwen3.8-flash-next-nvfp4-tp2 Public

    Python