Agent Harness Engineer

Nickel; San Francisco (CA); On-site; USD 180,000 - 260,000; Full time

Job summary

Nickel is hiring for a role at the forefront of AI infrastructure, building a proprietary Agent Harness stack that blends algorithms, engineering, and modeling. The team designs runtimes, memory, tools, and scheduling for ultra-long-horizon multi-agent tasks and ensures ongoing co-evolution with model training.

Ideal candidates combine deep expertise in LLMs, agent systems, and prompt/context engineering, with strong communication in English.

Qualifications

• Advanced knowledge of large language models and agent-based systems.

Responsibilities

• Research and implement cutting-edge Harness mechanisms.
• Manage context, memory, and tool orchestration for multi-agent tasks.
• Coordinate with model training to evolve Harness algorithms.
• Design system architecture for Agent runtime, scheduling, and sandboxing.

Skills

Deep learning

LLM tooling

Agent systems

Programming

Education

Master’s degree in CS or related field

Tools

Claude Code

Cursor

Codex

Copilot

Manus

OpenClaw

Job description

Company Description

Nickel is building AI infrastructure to enable doctors to deliver personalized, diagnostics-led longevity medicine to people everywhere. Backed by an eight-figure funding round, Nickel is preparing for launch and expanding its team with exceptional software and AI engineering talent. The company focuses on turning complex medical diagnostics into actionable insights that support better long-term health outcomes. Team members join a fast-paced, mission-driven environment with the opportunity to shape core infrastructure at the intersection of healthcare and artificial intelligence.

Role Description

Build Nickel’s proprietary Agent Harness technical stack through the synergy of algorithms, engineering, and modeling:

• Models handle reasoning and generation.
• Harness manages the entire runtime beyond the model: context, memory, tools, control loops, multi-agent scheduling, and self-evolution.
• Establish and close loop with the Model Training effort to drive the co-evolution of models and the Harness.

Key Responsibilities (Algorithms/Engineering Combined)

• Research and Implementation of Cutting-edge Harness Mechanisms
• Context Management & Context Engineering: Truncation, compression, KV Cache/prefix cache reuse, and cross-turn context strategies.
• Long-term Memory: Episodic/semantic memory, vector stores, session persistence, and replay.
• Agent Loop & Tooling: Agent Loop, Tool Use, Skills, MCP protocol integration, and tool orchestration.
• Multi-Agent Systems: Subagent communication, task decomposition, result aggregation, and hierarchical planning.
• Self-Evolving Agents: Execution and recovery of ultra-long-horizon tasks (multi-step, multi-day, multi-session).
• Model–Harness Synergy
• Define model requirements from a Harness perspective (e.g., reasoning formats, tool-call stability, planning signals).
• Surface failure modes from real-world tasks (e.g., planning collapse, tool misuse, context loss) to the training effort and work out solution.
• Evaluation and Research Loop
• Design Harness-level benchmarks, evaluation metrics, and data annotation strategies.
• Iterate through experiments using real-world tasks (internal and external scenarios) rather than relying solely on model benchmark scores.
• System Architecture and Engineering Implementation
• Participate in technical selection and architecture design for Harness products/systems (Agent runtime, scheduling layer, MCP client/server, sandboxing, state machines).
• Translate research prototypes into runnable, observable, and iterable production systems.
• Solve production-grade challenges such as interrupt recovery, streaming, long-running tasks, error fallback, and concurrent sub-agent management.

Qualifications

• Background in Computer Science or a related field. Master’s degree or above for the algorithm track (exceptional candidates may be exempted); bachelor’s degree or above from a reputable university for the engineering track.
• 2 years of research experience for the algorithm track (exceptional candidates may be exempted). For the engineering track: strong technical skills, broad vision, and the ability to independently drive modules.
• Deep user of code-focused and general-purpose Agent products (e.g., Claude Code, Cursor, Codex, Copilot, Manus, OpenClaw), integrating Agents into daily workflows.
• Proficiency in LLM and Agent fundamentals: LLM APIs, KV Cache, Agent Loop, Tool Use, Reasoning, Planning, Skills, MCP, Memory, Subagent, and Multi-Agent systems.
• Understanding & hands-on capability of model training workflows (pretraining / SFT / RLHF / DPO, etc.), with the ability to contribute to training data construction, preference labeling, and evaluation loop design.
• Capable of translating Harness-layer observations into interpretable, quantifiable improvement proposals for model training.
• Practical, hands-on understanding of Prompt Engineering, Context Engineering, and Harness Engineering (deep understanding for the research track; solid understanding for the engineering track).
• Strong intuition and judgment regarding model behavior; ability to distinguish between “model defects” and “Harness failure to handle edge cases.”
• Clear communication skills in both Chinese and English. For the algorithm track: emphasis on independent topic selection and 0-to-1 experimental iteration. For the engineering track: emphasis on system design and delivery quality.

Preferred Qualifications (Research/Engineering)

• Research or industry experience in AI, with publications in top-tier conferences or proven system deployment cases.
• Open-source projects or significant contributions to the open-source community.
• Experience in fast-paced product iteration within small teams (OPC / solo-founder experience).
• Engineers experienced in collaborating deeply with researchers, or researchers experienced in collaborating deeply with engineers (Research-Engineering fusion).
• Awards from competitions (ACM, Kaggle, SWE-bench, Mathematical Modeling, etc.).
• Proficiency in English reading/writing to engage with open-source communities and user groups.
• Experience adapting to specific model features (prefix cache, long context windows, reasoning formats).

工作综述

以算法、工程、模型协同的方式，构建 Nickel 自有的Agent Harness技术体系：**

Key Responsibilities (Algorithms/Engineering Combined)

• 模型负责推理与生成
• Harness负责上下文、记忆、工具、循环、多智能体调度、自进化等“模型之外的全部运行时”
• 与模型训练工作相结合，推动模型, Harness共同进化
• 上下文管理与 Context Engineering（截断、压缩、KV Cache/prefix cache复用、跨轮上下文策略）
• 长期记忆（episodic / semantic / vector store / 会话持久化与回放）
• Agent Loop、Tool Use、Skills、MCP协议与工具编排
• Subagent / Multi-Agent通信、任务分解、结果聚合、层级规划
• 自进化 Agent、超长程任务（多步、多天、多会话）执行与恢复
• 模型–Harness协同
• 从Harness视角定义模型需要的能力（推理格式、tool-call稳定性、planning信号等）
• 将真实任务中的失败模式（规划崩坏、工具误用、上下文丢失）回流给训练侧
• 评测与研究闭环
• 提出Harness层benchmark、评测维度、数据标注策略
• 用真实世界任务（内部任务 外部场景）做实验迭代，而非只看模型分数
• 系统架构与工程实现
• 参与Harness产品 / 系统的技术选型与架构设计（Agent runtime、调度层、MCP client/server、沙箱、状态机）
• 将研究原型转化为可运行、可观测、可迭代的工程系统
• 处理中断恢复、流式、长任务、错误回退、并发子Agent等生产级问题

任职要求

• 计算机或相关方向背景；算法侧硕士及以上（极优秀可放宽），工程侧知名高校本科及以上
• 算法侧 2年以上科研经验（优秀可放宽），工程侧技术硬、眼界广、能独立推进模块
• 深度使用代码类 通用类 Agent产品（Claude Code / Cursor / Codex / Copilot / Manus / OpenClaw 等），把 Agent 融进日常工作流
• 熟悉 LLM 与 Agent 基础机制：LLM API / KV Cache / Agent Loop / Tool Use / Reasoning / Planning / Skills / MCP / Memory / Subagent / Multi-Agent
  • 对 Prompt Engineering / Context Engineering / Harness Engineering 有实操级理解（算法侧深入理解，工程侧较深入）
• 理解并掌握模型训练流程（预训练 / SFT / RLHF / DPO 等），能参与训练数据构造、偏好标注与评测闭环设计
• 能基于Harness层观测结果，为模型训练提供可解释、可量化的改进
• 对模型行为有品味和判断力，能区分“模型缺陷”还是“harness没兜住”
• 中/英文沟通清晰；算法侧强调独立选题与0→1实验迭代能力，工程侧强调系统设计与交付质量

加分项（研究/工程通用）

• AI领域科研/工业经验，顶会论文或系统落地案例
• 开源作品、开源社区深度贡献
• 小团队快速迭代产品的经验（OPC / 一人公司实践）
• 与研究员深度协作的工程师，或与工程深度协作的研究员（研产融合）
• 比赛奖项（ACM / Kaggle / SWE-bench类/数学建模等）
• 英文读写可对接开源社区与用户社群
• 对模型特性（prefix cache、长上下文、推理格式）有适配经验