ModelBest
Posted 1y ago

大模型算法工程师-智能体方向

ModelBest
Beijing, Beijing, China
OnsiteFull Time
Responsibilities
  • building systems
  • designing harness
  • evaluating models
Requirements
  • Master's degree in CS/AI
  • 1+ years large-model experience
  • Proficient Python
  • Knowledge of Java/C++
  • Experience with RAG/Agent frameworks
  • Prompt engineering
  • Evaluation and deployment
Technical tools mentioned
PythonJavaC++Retrieval-Augmented Generation (RAG)LoRASFTDPOBM25

Job description

1. 端到端智能系统构建与优化:覆盖意图识别、多轮对话改写、知识检索、内容生成与用户偏好对齐全链路,持续提升交互质量与系统稳定性;
2. Agent Harness 设计与建设:设计并实现稳定可扩展的智能体运行框架(Agent Harness),负责工具调用编排、上下文与记忆管理、执行循环控制、错误恢复与重试、沙箱化执行环境等核心能力,为上层 Agent 提供可靠的运行底座;
3. Agent 能力与范式演进:实现具备任务规划、工具调用、自我反思能力的 Agent;探索规划执行、反思迭代等范式的工程落地,以及多 Agent 协作机制在复杂任务分解中的应用;推动 Agent 长程任务执行的稳定性与可观测性建设;
4. 垂直领域大模型应用落地:参与教育等场景对话式 AI 产品的全流程研发,探索长上下文、结构化推理与多模态能力在真实业务中的创新应用;
5. 模型能力与评测体系:结合业务数据开展指令微调、偏好对齐及 Prompt 工程;推动 Agent 轨迹(trajectory)评测、工具调用成功率、端到端任务完成率等指标体系建设,构建覆盖准确性、可靠性与安全性的自动化评测框架。

1. 学历与经验:计算机、人工智能或相关专业硕士及以上学历,具备 1 年以上大模型相关工作经验;
2. 编程能力:熟练掌握 Python,了解 Java 或 C++ 等语言;具备良好的工程素养,能独立完成模型服务部署与接口开发;
3. Agent 工程能力:理解 Agent Harness 的核心组成(执行循环、工具调用协议、上下文窗口管理、状态与记忆、超时与重试、安全沙箱等),具备主流 RAG/Agent 开发框架的实战经验,有从 0 到 1 完整交付项目的经历;
4. 核心技术掌握:深入理解 RAG 架构(混合检索、重排序、Chunking 策略)、Agent 工作机制、Prompt 工程、检索算法(倒排索引、语义检索、BM25),以及主流微调与对齐技术(SFT、LoRA、DPO);
5. 系统工程能力:了解大模型推理优化(量化、KV Cache 复用等)和向量数据库的工程实践;熟悉异步并发、流式输出、可观测性(日志/追踪/监控)等服务端工程;有 LLMOps 工具链使用经验者优先。

加分项(满足其一即可):
- 在 ACL、EMNLP、ICLR、NeurIPS 等顶级会议或期刊发表过相关论文
- 有从零搭建或深度优化 Agent Harness / 智能体运行框架的经验
- 有大规模 Agent 系统或多智能体协作项目的完整落地与迭代经验
- 熟悉 Agent 工具调用标准化协议,具备自定义工具链开发能力
- 具备推理增强模型(链式思维、慢思考)或长上下文场景的工程经验
- 参与过开源大模型 / Agent 框架社区贡献,或有相关赛事获奖经历

About ModelBest

Develops efficient, on-device large language models and edge AI.

Year founded
2022
Employees
200
Organization type
Private
Latest investment
Raised $5.00B Funding Round (2026) — led by China Telecom, Shenzhen Capital Group, Inovance Capital, Primavera Capital Group
Subsidiaries
Headquarters
CN

Similar jobs

Large Model Algorithm Engineer roles near Beijing, Beijing
1d
Save
Mark Applied
Hide
大模型算法工程师(语音) - 火山方舟
Beijing or Shenzhen or Shanghai or Hangzhou
OnsiteFull Time
ByteDance
ByteDance: Developing AI-driven content platforms and mobile applications.
Master's degree or higher graduating in 2027; relevant computer or electronic information studies preferred. Strong C/C++ or Python, algorithms, data structures, and AI research skills preferred.
C, C++, Python, NeurIPS, ICML, ICLR, CVPR, ICCV, ACL, KDD, ACM/ICPC, NOI/IOI, Top Coder, Kaggle, SeedTTS, CosyVoice, Qwen-omni, Veo3
2w
Save
Mark Applied
Hide
微信搜索-LLM大模型算法工程师
Beijing, Beijing, China
OnsiteFull Time
Tencent
TencentHKEX: 0700: Provides integrated internet services, digital entertainment, and cloud technology.
4+ YOEDeep understanding of large language models and experience in AI Agent, RAG, RL/RLHF; experience in AI search, multimodal Q&A or deep research preferred; strong product sense; publications/industry LLM experience a plus.
2w
Save
Mark Applied
Hide
【27届校招】大模型算法工程师(智驾/机器人方向)
Beijing or Guangzhou or Shenzhen or Shanghai
OnsiteFull Time
XPeng
XPengNYSE: XPEV: Designs and manufactures smart electric vehicles and autonomous technology.
Campus hire (class of 2027) in CS/AI/Electronic Engineering/Automation/Math; strong Python skills, C++ preferred; experience with large models, SFT/RL, model training, data curation, and problem solving.
Python, C++
3w
Save
Mark Applied
Hide
大模型算法工程师(资管)-2027届
Beijing or Chengdu or Chongqing
OnsiteFull Time
Mashang Consumer Finance
Mashang Consumer Finance: A licensed financial institution providing digital consumer lending services.
Master's in CS or related preferred, strong programming (Python/C/C++), Linux experience, background in ML/NLP/IR/AI, familiarity with large models, strong learning and communication skills.
python, C, C++, linux
3w
Save
Mark Applied
Hide
L4智驾大模型团队-端到端大模型算法工程师(Mstar)
Beijing or Shanghai
OnsiteFull Time
Momenta
MomentaHong Kong Stock Exchange: 06880: Develops autonomous driving software and intelligent transportation solutions.
Master's degree in CS/Automation/AI (211/985) or higher; deep knowledge of Transformer/LLM/VLM and generative models; strong Python/C++ skills; experience with on-board deployment (TensorRT/CUDA) and planning/prediction with imitation/offline RL.
Transformer, State-space Model, Mamba, Diffusion Model, VAE, GAN, LLM, VLM, Python, C++, TensorRT, CUDA, IL, RL, Auto-labeling
1mo
Save
Mark Applied
Hide
提前批-大模型算法工程师
Wuhan or Hefei or Shanghai or Beijing
OnsiteFull Time
Nio
NioNYSE: NIO: Designs and manufactures smart premium electric vehicles.
Master's degree or higher in CS/math/stat preferred; strong algorithms experience (time series, CTR, ranking, causal inference); model development, deployment and optimization experience; strong communication and cross-team collaboration.
1mo
Save
Mark Applied
Hide
基础大模型算法工程师- DLLM
Beijing or Shanghai
OnsiteFull Time
Xiaohongshu
Xiaohongshu: Social media and e-commerce platform for lifestyle sharing.
Bachelor's degree in CS/AI/math or higher; strong experience with diffusion and large-model training, data processing, model parallelism, and ML research; proficient in Python/C++/Java.
Python, C++, Java
1mo
Save
Mark Applied
Hide
大模型算法工程师 / 专家
Beijing or Shanghai
OnsiteFull Time
XPeng
XPengNYSE: XPEV: Chinese electric vehicle manufacturer developing smart mobility technology.
Master's or higher in CS/AI/Automation/Math, strong deep learning and reinforcement learning background, experience with large models, PyTorch, and multi-sensor perception for autonomous driving.
PyTorch, PPO, DPO, GRPO, SAC, Agent, Harness