ModelBest

大模型工程师-推理与部署优化

ModelBest  •  Onsite  •  21 hours ago
Apply
AI can make mistakes so check important info. Chat history is never stored.

Job Description

大模型工程师-推理与部署优化
北京
社招
全职
技术 - 基础架构
职位 ID:A23292
职位描述
1、负责大语言模型、多模态模型的推理引擎搭建、优化与线上落地,支撑高并发、低延迟的模型服务;
2、基于 vLLM、TensorRT-LLM、SGLang、LightLLM 等主流推理框架进行性能调优,并探索 推测性解码 (Speculative Decoding)、前缀缓存 (Prefix Caching) 等前沿加速技术;
3、负责大模型在 NVIDIA A/H 系列、昇腾 等国产加速卡上的适配与算子优化,解决兼容性与性能瓶颈;
4、设计与实现高性能模型 Serving 架构,包括 Prefill-Decode (PD) 分离架构、Continuous Batching (连续批处理)、负载均衡、流式输出等;
5、基于 Docker、Kubernetes 实现模型服务容器化编排、弹性扩缩容与运维稳定性保障;
6、定位并解决线上服务瓶颈:如显存碎片化、OOM、推理崩溃、时延抖动(Tail Latency)、多卡并行异常等;
7、与算法团队协作,将训练好的模型权重标准化、工程化,实现快速上线与迭代。
职位要求
1、本科及以上学历,计算机相关专业,2 年及以上大模型部署或推理优化相关经验;
2、熟悉多种主流大模型推理框架(如 vLLM, SGLang, TensorRT-LLM),深入理解其 PagedAttention、KV Cache 管理等核心机制;
3、熟悉 CUDA 或昇腾 CANN 开发,有算子级优化经验者优先;
4、具备模型服务化与高并发架构设计经验,熟悉流式推理、流量管控等策略;
5、熟练使用 Nsight Systems, PyTorch Profiler 等性能分析工具进行瓶颈分析;
6、熟悉 Linux 环境,具备较强的问题排查能力,能独立定位崩溃、性能瓶颈、硬件兼容性问题;
7、有在 A100/H100 / 昇腾 910B 等硬件上部署优化百亿 / 千亿参数大模型经验者优先。
加分项:
1、有国产硬件(昇腾 / 寒武纪 / 海光等)适配与优化经验;
2、对 SGLang、vLLM 等框架有源码级修改或核心贡献;
3、具备 PD 分离 或 集群算力调度优化 经验;
4、熟悉监控告警、服务可观测性体系。
投递
ModelBest

About ModelBest

ModelBest established in August 2022, is an Artificial Intelligence technology company headquartered in Beijing, China. The company is deeply rooted in the field of general AI, with a focus on the innovation and application transformation of large-scale models. ModelBest boasts a prestigious founding R&D team from the Tsinghua in the AI field. Leveraging numerous cutting-edge technologies in natural language processing, the company is currently constructing a large-scale pre-trained model library and corresponding tools, aiming to standardize the technology and applications of large models.

Based on the capabilities of the CPM series of large models, ModelBest has successfully facilitated the intelligent upgrade and efficiency enhancement of various industries. Moreover, the company has released a multi-modal dialog assistant with hundreds of billions of parameters, “Luca", to the public. Adhering to the core philosophy of "intelligence encompassing all", we look forward to exploring and expanding unknown possibilities with global partners, aiming to ensure AI technology serves humanity for a better life in a safe and inclusive manner, laying a solid foundation for the advent of the AGI world.

OpenBMB (Open Lab for Big Model Base,https://github.com/OpenBMB), founded by ModelBest Inc & TsinghuaNLP, aims to build foundation models and systems towards AGI.

Industry
Biotech & Life Sciences
Company Size
11-50 employees
Headquarters
Beijing, CN
Year Founded
2022
Social Media