米哈游大模型平台开发工程师
社招全职程序&技术类地点:上海状态:招聘
工作描述
Qualifications 必须满足(硬性): - 计算机相关本科+,3 年+ 模型平台/推理优化/MLOps(硕士 2 年+) - 实操过量化与格式转换:FP8/INT8/INT4(GPTQ/AWQ),HF → TRT-LLM/vLLM/SGLang/ONNX,并能做转换后数值/精度验证 - 实操过精度评测与压测:搭建过评测集/流水线,用过 lm-eval-harness / vLLM bench / TRT-LLM benchmark,理解 TTFT/TPOT/吞吐/MFU - 有模型上线准入的 CI/CD 自动化流水线经验(Argo/Airflow/Jenkins/GitLab CI 任一) - 熟悉 K8s,有基于 K8s 的任务/流水线运行与排障 - 熟悉至少一种推理引擎(vLLM/TRT-LLM/SGLang/TGI)原理与使用,懂 GPU(CUDA/显存/并行) - 熟练 Python(必须),了解 C++/Go,有 PyTorch/HF Transformers 经验 加分项 - 有多算力卡适配与评测经验(NVIDIA + AMD/昇腾/寒武纪/海光,会其一即可,越多越好) - 做过统一推理后端适配层或跨卡 CI 矩阵 - 有成本建模(…
登录查看完整工作描述
微信扫码,1秒登录
包括英文材料
大模型+
https://www.youtube.com/watch?v=xZDB1naRUlk
You will build projects with LLMs that will enable you to create dynamic interfaces, interact with vast amounts of text data, and even empower LLMs with the capability to browse the internet for research papers.
https://www.youtube.com/watch?v=zjkBMFhNj_g
vLLM+
https://www.newline.co/@zaoyang/ultimate-guide-to-vllm--aad8b65d
vLLM is a framework designed to make large language models faster, more efficient, and better suited for production environments.
https://www.youtube.com/watch?v=Ju2FrqIrdx0
vLLM is a cutting-edge serving engine designed for large language models (LLMs), offering unparalleled performance and efficiency for AI-driven applications.
SGLang+
[英文] Install SGLang
https://docs.sglang.ai/get_started/install.html
SGLang is a fast serving framework for large language models and vision language models.
https://github.com/sgl-project/sgl-learning-materials
ONNX+
https://github.com/onnx/tutorials
Open Neural Network Exchange (ONNX) is an open standard format for representing machine learning models.
[英文] Introduction to ONNX
https://onnx.ai/onnx/intro/
This documentation describes the ONNX concepts (Open Neural Network Exchange).
CI+
https://www.ibm.com/cn-zh/think/topics/continuous-integration
持续集成 (CI) 是一种软件开发实践,开发人员在整个开发周期中会定期将新的代码和代码变更集成到中央代码存储库中。它是 DevOps 和敏捷方法的关键组成部分。
https://www.youtube.com/watch?v=42UP1fxi2SY
CD+
https://www.redhat.com/zh-cn/topics/devops/what-is-ci-cd
CI/CD 是持续集成和持续交付/部署的缩写,旨在简化并加快软件开发生命周期。
https://www.youtube.com/watch?v=R8_veQiYBjI&list=PLy7NrYWoggjzSIlwxeBbcgfAdYoxCIrM2
Argo+
https://argo-cd.readthedocs.io/en/stable/understand_the_basics/
Before effectively using Argo CD, it is necessary to understand the underlying technology that the platform is built on.
https://www.youtube.com/watch?v=MeU5_k9ssrs
The ArgoCD chapter includes building a pipeline of dynamically updating & building a new application version using GitLab downstream pipeline feature.
Airflow+
[英文] Tutorials - Airflow
https://airflow.apache.org/docs/apache-airflow/stable/tutorial/index.html
Once you have Airflow up and running with the Quick Start, these tutorials are a great way to get a sense for how Airflow works.
https://www.youtube.com/watch?v=K9AnJ9_ZAXE
In this 2-hour Airflow Tutorial for Beginners Full Course, we combine theory explanation and practical demos to help you get started quickly as an absolute beginner.
Jenkins+
https://www.youtube.com/watch?v=f4idgaq2VqA
GitLab+
https://docs.gitlab.com/tutorials/
Learn about GitLab fundamentals by following guided instructions.
Kubernetes+
https://kubernetes.io/docs/tutorials/kubernetes-basics/
This tutorial provides a walkthrough of the basics of the Kubernetes cluster orchestration system.
https://kubernetes.io/zh-cn/docs/tutorials/kubernetes-basics/
本教程介绍 Kubernetes 集群编排系统的基础知识。每个模块包含关于 Kubernetes 主要特性和概念的一些背景信息,还包括一个在线教程供你学习。
https://www.youtube.com/watch?v=s_o8dwzRlu4
Hands-On Kubernetes Tutorial | Learn Kubernetes in 1 Hour - Kubernetes Course for Beginners
https://www.youtube.com/watch?v=X48VuDVv0do
Full Kubernetes Tutorial | Kubernetes Course | Hands-on course with a lot of demos
推理引擎+
https://www.youtube.com/watch?v=_dvk75LEJ34
https://www.youtube.com/watch?v=XtT5i0ZeHHE
还有更多 •••
相关职位

社招测试开发
1. 本科及以上学历,计算机/软件工程等专业背景 2. 熟练掌握 Python 或 Go,能独立编写测试脚本或工具 3. 熟悉 Linux,了解常见测试类型(功能、接口、性能、稳定性) 4. 熟悉至少
更新于 2025-12-10北京|杭州|上海

社招3年以上技术类-开发
1、3年以上JAVA开发经验,理解io、多线程、集合等基础框架,了解JVM原理,有良好的编程习惯; 2、熟悉分布式系统设计,熟悉分布式、缓存、消息等机制;能对分布式常用技术进行合理应用,解决问题; 3
更新于 2026-04-03北京
社招5年以上程序&技术类
- 计算机科学与技术、软件工程、人工智能等相关专业,本科及以上学历 - 5 年以上分布式训练系统或 K8s 平台研发经验 - **K8s 调度**:精通 kube-scheduler framewor
上海|北京