高级机器学习工程师 (代理系统)
Senior Machine Learning Engineer (Agent Systems)
关于 e:cue
e:cue 是一家快节奏、高增长的初创公司，为营销、财务和收入领域的领导者打造定制化 AI 分析师。我们的平台结合了生产级应用服务、云基础设施和代理系统，支持高风险业务决策。我们正在寻找一位高级工程师，能够从工单到结果全程负责：规划、构建、交付，并在生产环境中负责维护。
职位概述
该职位负责代理堆栈的核心部分，决定代理如何规划和执行，如何与数据交互，以及如何随时间评估和改进它们。
你将参与以下领域：
- 代理系统：规划、工具使用、多代理编排、长上下文工作流
- 后端和基础设施：代理服务、数据管道和可观测性
- 评估和训练后：设计评估框架、反馈循环、数据集，并改进代理行为
你将负责的工作
设计和构建生产级代理系统：
- 工具执行框架（MCP 服务器、沙盒环境、工具架构）
- 规划和推理流水线
- 上下文和依赖关系感知的代理执行
负责支撑生产代理的服务：
- 可靠性、延迟和扩展性的改进
- 可观测性完整性（日志、追踪、离线和在线评估的评估钩子）
开发评估和反馈系统：
- 定义代理性能指标（离线和在线）
- 负责评估框架和测试套件
- 在系统中添加 instrumentation 以生成高质量的评估和训练数据
参与训练后的模型改进：
- 数据集生成（轨迹收集、偏好数据）
- 模块微调（SFT、DPO 等），当上下文工程不足以满足需求时
- 提示和系统设计，以提升推理和上下文管理能力
我们寻找的人选
- 有构建或使用 LLM 驱动系统的经验
- 熟悉代理、工具使用或结构化推理系统
- 有处理模糊目标的 ML 评估系统的经验
- 能够从头到尾负责问题
- 强烈的产品直觉
加分项
- 有 ML 系统或训练工作流的经验，包括微调（SFT、DPO、RLHF 等）、数据集构建和评估流水线
- 有为使用工具的 LLM 构建代理框架的经验，适用于长上下文或检索密集型工作流
- 熟悉现代推理、前沿 API 和部署栈（vLLM、SGLang 或类似）
- 在初创公司中独立负责大型系统的经验
查看英文原文
About e:cue
e:cue is a fast-paced, high-growth startup building custom AI analysts for leaders in marketing, finance, and revenue. Our platform combines production-grade application services, cloud infrastructure, and agent systems that power high-stakes business decisions. We're looking for a senior engineer who can take work from ticket to outcome: scope it, build it, ship it, and own it in production.
Role Overview
This role owns core parts of the agent stack, deciding how agents plan and execute, how they interact with data, and how we evaluate and improve them over time.
You'll work across:
- Agent systems: planning, tool use, multi agent orchestration, long-context workflows
- Backend and infrastructure: agent services, data pipelines, and observability
- Evaluation and post-training: Designing evaluation harnesses, feedback loops, datasets, and improving agent behavior
What You'll Do
Design and build production agent systems:
- Tool execution frameworks (MCP servers, sandbox environments, tool architectures)
- Planning and reasoning pipelines
- Context and dependency aware agent execution
Own services that power production agents:
- Reliability, latency, and scaling improvements
- Observability integrity (logging, tracing, evaluation hooks for offline and online evaluation)
Develop evaluation and feedback systems:
- Define metrics for agent performance (offline and online)
- Own evaluation harnesses and test suites
- Instrument systems to generate high-quality evaluation and training data
Contribute to post-training and model improvement:
- Dataset generation (trajectory collection, preference data)
- Fine-tuning (SFT, DPO, etc.) for modules where context engineering isn't enough
- Prompting and system design for better reasoning and context management
What We're Looking For
- Experience building or working with LLM-powered systems
- Familiarity with Agents, tool use, or structured reasoning systems
- Experience with ML evaluation systems for ambiguous objectives
- Ability to own problems end-to-end
- Strong product intuition
Nice to Have
- Experience with ML systems or training workflows, finetuning (SFT, DPO, RLHF, etc.), dataset construction and evaluation pipelines
- Experience building agent frameworks for tool-using LLMs for long-context or retrieval-heavy workflows
- Familiarity with modern inference, frontier APIs, and serving stacks (vLLM, SGLang, or similar)
- Experience at a startup owning large systems independently
Why Join Us
- Build meaningful AI capabilities with direct business impact.
- Own state of the art agent development for domain specific workflows
- Work on both product and infrastructure with real ownership.
- Help shape architecture, reliability standards, and team practices as we scale.
- Join a small team where strong execution is visible and valued.
Our Benefits and Total Rewards
- Remote team: work from where you need to
- Flexible paid time off: because you're an adult
- Generous health insurance reimbursement through QSEHRA
- Competitive salary
- Equity packages
- Company-performance bonus
Equal Opportunity Employer
We are an equal opportunity employer and value diversity at our company. We do not discriminate on the basis of race, religion, color, national origin, gender, gender identity or expression, sexual orientation, age, marital status, veteran status, disability status, or any other legally protected characteristic. We are committed to creating an inclusive environment for all employees and encourage individuals from underrepresented backgrounds to apply.
Originally posted on Himalayas