🌏 全球远程工作机会汇总

在招 0 仅中国地区远程 0 👁 累计浏览 1072 更新时间

首席软件工程师——可观测性与遥测数据

Principal Software Engineer - Observability & Telemetry Data

全球可投 开发

🏢 公司 / 雇主

Smartsheet

发布于 — · 首次收录 2026-11-16 · 最后确认在线 2026-11-16 · 👁 浏览 2 次

🛠 技能要求

OpenTelemetryDatabricksMLflowDelta LakeApache IcebergApache SparkFluentBitAmazon KinesisDatadog分布式追踪RUM

📋 职位要求

  • 设计遥测数据平台架构:负责遥测数据落地、建模与查询方式的端到端设计,包括开放表格式、分区与模式演进策略、存储与计算分离、分层保留,以及工程师实际使用的查询接口;使遥测成为持久、可移植、由 Smartsheet 自有的数据集,而非受供应商锁定的副产品。 Architect the Telemetry Data Platform: Own the end-to-end design for how telemetry is landed, modeled, and queried, including open table formats, partitioning and schema evolution strategy, separation of storage and compute, tiered retention, and the query interfaces engineers actually use. Make telemetry a durable, portable, Smartsheet-owned dataset rather than a vendor-locked byproduct.
  • 掌控遥测数据模型:定义语义约定、共享业务标识符(用户、组织、套餐、租户)以及使任意信号都能与其他信号相互关联的模式标准,并推动 Smartsheet 每个服务团队采用这些标准。 Own the Telemetry Data Model: Define the semantic conventions, shared business identifiers (user, org, plan, tenant), and schema standards that let any signal be correlated with any other, and drive their adoption across every service team at Smartsheet.
  • 制定 OpenTelemetry 方向:主导向基于 OTel 的埋点迁移,定义采集器架构、上下文传播与采样策略(包括尾部采样),让工程师能够从一条日志追到一条追踪再到一个指标而不丢失脉络。 Set OpenTelemetry Direction: Lead the migration to OTel-based instrumentation, defining collector architecture, context propagation, and sampling strategy (including tail-based sampling) so that engineers can move from a log line to a trace to a metric without losing the thread.
  • 将 AI 与智能体遥测集成到 Databricks:负责将我们的可观测性平台连接到 Databricks 和 MLflow 的架构,使智能体与模型遥测(提示词、补全、工具与 MCP 调用、评估结果)以不采样方式捕获,在我们治理所要求的输入/输出脱敏条件下依然有用,并能与主可观测性技术栈中的追踪和指标干净地对账。 Integrate AI and Agentic Telemetry with Databricks: Own the architecture connecting our observability platform to Databricks and MLflow, so that agentic and model telemetry (prompt, completion, tool and MCP calls, evaluation results) is captured un-sampled, stays useful under the input/output redaction our governance requires, and reconciles cleanly with the traces and metrics in our primary observability stack.
  • 为数据平台本身做埋点:为我们的数据资产(包括 Databricks 作业、管道和数据仓库)带来一流的可观测性,提供关于新鲜度、数据质量、血缘和成本的有效信号,让数据可靠性得到与服务可靠性同样严格的度量。 Instrument the Data Platform Itself: Bring first-class observability to our data estate, including Databricks jobs, pipelines, and warehouses, with meaningful signals for freshness, data quality, lineage, and cost, so data reliability is measured with the same discipline as service reliability.
  • 大规模设计采集与路由:设计高吞吐量的采集与路由层(FluentBit、Kinesis 和 OTel 采集器),将遥测数据从每个服务跨美国、欧盟、澳大利亚和 GovCloud 区域传送到目的地,并负责这些管道在整合到统一后端过程中的迁移。 Engineer Collection and Routing at Scale: Architect the high-volume collection and routing tier (FluentBit, Kinesis, and OTel collectors) that moves telemetry from every service to its destination across US, EU, AU, and GovCloud regions, and own the migration of these pipelines as we consolidate onto a unified backend.
  • 掌控遥测经济学:为可观测性数据设定成本架构,包括摄取治理、基数控制和存储分层,让团队获得调试所需的保真度,同时不因支出压力而减少埋点。 Own Telemetry Economics: Set the cost architecture for observability data, including ingest governance, cardinality control, and storage tiering, so that teams get the fidelity they need to debug without the spend pressure that causes them to under-instrument.
  • 跨组织边界推行架构:与数据平台、AI 平台和基础设施组织合作,使遥测架构与他们的架构保持一致,影响并非由您掌管的路线图,并在公司层面的平台与供应商决策中代表可观测性的立场。 Carry Architecture Across Org Boundaries: Partner with the Data Platform, AI Platform, and infrastructure organizations to align telemetry architecture with theirs, influence roadmaps you do not own, and represent observability in company-level platform and vendor decisions.
  • 提升技术标准:主持设计与代码评审,撰写他人据此构建的架构决策与标准,并在埋点、遥测数据建模和成本感知设计方面指导资深与中级工程师。 Raise the Technical Bar: Lead design and code reviews, author the architecture decisions and standards others build against, and mentor senior and mid-level engineers on instrumentation, telemetry data modeling, and cost-aware design.
  • 10 年以上构建和运营大规模分布式系统、数据平台或可观测性基础设施的经验,并有 Principal 或 Staff 级别的任职经历。 10+ years of experience building and operating large-scale distributed systems, data platforms, or observability infrastructure, including time at Principal or Staff level.
  • 深厚的可观测性专长:在大规模场景下对指标、日志和分布式追踪拥有实际生产运维经验,包括至少一种主流后端(Datadog 或同级产品),并对基数与成本机制有切实的理解。 Deep Observability Expertise: Hands-on production ownership of metrics, logs, and distributed tracing at scale, including at least one major backend (Datadog, or comparable) and a working understanding of cardinality and cost mechanics.
  • 遥测数据工程:在大规模数据架构方面展现出深厚造诣,包括开放表格式(Delta Lake、Apache Iceberg)、Spark 或同级分布式处理、流式摄取、分区与模式演进,以及对超大数据集的查询性能与成本调优。 Telemetry Data Engineering: Demonstrated depth in large-scale data architecture, including open table formats (Delta Lake, Apache Iceberg), Spark or comparable distributed processing, streaming ingestion, partitioning and schema evolution, and query performance and cost tuning over very large datasets.

📄 职位描述(中文)

二十多年来,Smartsheet 一直在帮助团队无缝地管理工作、更聪明地扩展解决方案。如今,在我们迄今最雄心勃勃的篇章中,我们正将人类团队与 AI 智能体联合起来。通过编排智能体最擅长的工作、将手动任务自动化,并大规模发掘洞察,我们为人们腾出空间,使其专注于真正重要的事情:判断力、创造力和宏观思考。这就是工作中的魔法,也是我们每天为之奋斗的目标。 可观测性工程团队全面负责 Smartsheet 可观测性与遥测工作的广度与深度。我们构建并运营一个每个工程团队都依赖的共享平台,帮助他们了解自己的软件在生产环境中的运行表现(指标、日志、分布式追踪、RUM、合成监控以及 AI/智能体遥测),覆盖全球多区域、商业与政府业务版图。我们正在把所有这些能力整合为一个统一的关联平台,并将遥测数据当作它本来的样子来对待:公司规模最大、最具价值的数据集之一。这是一项具有真正塑造空间的广泛职责,我们对这一内部平台秉持与面向客户的软件相同的严谨性、可靠性和产品标准。 作为首席软件工程师(可观测性与遥测数据),您将为 Smartsheet 收集、建模、存储遥测数据并从中提炼价值的方式设定技术方向。这是一个架构师角色:您将定义遥测数据平台及其治理标准,将该架构推广到各工程组织边界之外,与我们的数据和 AI 平台团队合作进行 Databricks 和 MLflow 集成,并充当公司在可观测性领域最权威的技术声音。您将影响并非由您掌管的路线图,解决保真度与成本之间的艰难设计权衡,并为每一位为服务做埋点的工程师提升技术标准。 该全职职位向可观测性工程团队负责人汇报,可驻于我们位于华盛顿州贝尔维尤的办公室,也可以在美国境内 Smartsheet 为注册雇主的任何地点远程工作。 你将负责: 设计遥测数据平台架构:负责遥测数据落地、建模与查询方式的端到端设计,包括开放表格式、分区与模式演进策略、存储与计算分离、分层保留,以及工程师实际使用的查询接口;使遥测成为持久、可移植、由 Smartsheet 自有的数据集,而非受供应商锁定的副产品。 掌控遥测数据模型:定义语义约定、共享业务标识符(用户、组织、套餐、租户)以及使任意信号都能与其他信号相互关联的模式标准,并推动 Smartsheet 每个服务团队采用这些标准。 制定 OpenTelemetry 方向:主导向基于 OTel 的埋点迁移,定义采集器架构、上下文传播与采样策略(包括尾部采样),让工程师能够从一条日志追到一条追踪再到一个指标而不丢失脉络。 将 AI 与智能体遥测集成到 Databricks:负责将我们的可观测性平台连接到 Databricks 和 MLflow 的架构,使智能体与模型遥测(提示词、补全、工具与 MCP 调用、评估结果)以不采样方式捕获,在我们治理所要求的输入/输出脱敏条件下依然有用,并能与主可观测性技术栈中的追踪和指标干净地对账。 为数据平台本身做埋点:为我们的数据资产(包括 Databricks 作业、管道和数据仓库)带来一流的可观测性,提供关于新鲜度、数据质量、血缘和成本的有效信号,让数据可靠性得到与服务可靠性同样严格的度量。 在遥测之上构建分析层:将遥测转化为可供决策的分析,涵盖可靠性与事故指标、遥测成本与计费回收模型,以及管理层可据此采取行动的采用率与覆盖率报告。 大规模设计采集与路由:设计高吞吐量的采集与路由层(FluentBit、Kinesis 和 OTel 采集器),将遥测数据从每个服务跨美国、欧盟、澳大利亚和 GovCloud 区域传送到目的地,并负责这些管道在整合到统一后端过程中的迁移。 掌控遥测经济学:为可观测性数据设定成本架构,包括摄取治理、基数控制和存储分层,让团队获得调试所需的保真度,同时不因支出压力而减少埋点。 跨组织边界推行架构:与数据平台、AI 平台和基础设施组织合作,使遥测架构与他们的架构保持一致,影响并非由您掌管的路线图,并在公司层面的平台与供应商决策中代表可观测性的立场。 提升技术标准:主持设计与代码评审,撰写他人据此构建的架构决策与标准,并在埋点、遥测数据建模和成本感知设计方面指导资深与中级工程师。 参与生产支持和值班轮换,负责解决最复杂的问题,并推动能够提升系统韧性的根因分析。 你需具备: 10 年以上构建和运营大规模分布式系统、数据平台或可观测性基础设施的经验,并有 Principal 或 Staff 级别的任职经历。 深厚的可观测性专长:在大规模场景下对指标、日志和分布式追踪拥有实际生产运维经验,包括至少一种主流后端(Datadog 或同级产品),并对基数与成本机制有切实的理解。 遥测数据工程:在大规模数据架构方面展现出深厚造诣,包括开放表格式(Delta Lake、Apache Iceberg)、Spark 或同级分布式处理、流式摄取、分区与模式演进,以及对超大数据集的查询性能与成本调优。 Databricks 深度:在……方面的实践经验。
🇬🇧 查看英文原文(Original description)
For over 20 years, Smartsheet has empowered teams to manage work seamlessly and scale solutions smarter. Now, in our most ambitious chapter yet, we are uniting human teams with AI agents. By orchestrating the work agents do best, automating manual tasks and uncovering insights at scale, we create the space for people to focus on what truly matters: judgment, creativity, and big thinking. That is magic at work, and it’s what we show up for every day. Observability Engineering owns the breadth and depth of observability and telemetry at Smartsheet. We build and operate the shared platform every engineering team depends on to understand how their software behaves in production (metrics, logs, distributed traces, RUM, synthetics, and AI/agentic telemetry) across a global, multi-region, commercial and government footprint. We are consolidating ownership of all of it into a single correlated platform, and treating telemetry as what it actually is: one of the largest and most valuable datasets at the company. It's a broad mandate with real scope to shape, and we hold this internal platform to the same rigor, reliability, and product standards as customer-facing software. As a Principal Software Engineer (Observability & Telemetry Data) , you will set the technical direction for how Smartsheet collects, models, stores, and derives value from telemetry data. This is an architect role: you will define the telemetry data platform and the standards that govern it, carry that architecture across engineering org boundaries, partner with our Data and AI Platform teams on Databricks and MLflow integration, and act as the company's definitive technical voice on observability. You will influence roadmaps you do not own, resolve the hard design trade-offs between fidelity and cost, and raise the technical bar for every engineer instrumenting a service. This full-time position reports to the Team Lead, Observability Engineering and can be located in our Bellevue, WA office, or you may work remotely from anywhere in the US where Smartsheet is a registered employer. You Will Architect the Telemetry Data Platform: Own the end-to-end design for how telemetry is landed, modeled, and queried, including open table formats, partitioning and schema evolution strategy, separation of storage and compute, tiered retention, and the query interfaces engineers actually use. Make telemetry a durable, portable, Smartsheet-owned dataset rather than a vendor-locked byproduct. Own the Telemetry Data Model: Define the semantic conventions, shared business identifiers (user, org, plan, tenant), and schema standards that let any signal be correlated with any other, and drive their adoption across every service team at Smartsheet. Set OpenTelemetry Direction: Lead the migration to OTel-based instrumentation, defining collector architecture, context propagation, and sampling strategy (including tail-based sampling) so that engineers can move from a log line to a trace to a metric without losing the thread. Integrate AI and Agentic Telemetry with Databricks: Own the architecture connecting our observability platform to Databricks and MLflow, so that agentic and model telemetry (prompt, completion, tool and MCP calls, evaluation results) is captured un-sampled, stays useful under the input/output redaction our governance requires, and reconciles cleanly with the traces and metrics in our primary observability stack. Instrument the Data Platform Itself: Bring first-class observability to our data estate, including Databricks jobs, pipelines, and warehouses, with meaningful signals for freshness, data quality, lineage, and cost, so data reliability is measured with the same discipline as service reliability. Build the Analytics Layer on Telemetry: Turn telemetry into decision-grade analytics, covering reliability and incident metrics, telemetry cost and chargeback models, and adoption and coverage reporting that leadership can act on. Engineer Collection and Routing at Scale: Architect the high-volume collection and routing tier (FluentBit, Kinesis, and OTel collectors) that moves telemetry from every service to its destination across US, EU, AU, and GovCloud regions, and own the migration of these pipelines as we consolidate onto a unified backend. Own Telemetry Economics: Set the cost architecture for observability data, including ingest governance, cardinality control, and storage tiering, so that teams get the fidelity they need to debug without the spend pressure that causes them to under-instrument. Carry Architecture Across Org Boundaries: Partner with the Data Platform, AI Platform, and infrastructure organizations to align telemetry architecture with theirs, influence roadmaps you do not own, and represent observability in company-level platform and vendor decisions. Raise the Technical Bar: Lead design and code reviews, author the architecture decisions and standards others build against, and mentor senior and mid-level engineers on instrumentation, telemetry data modeling, and cost-aware design. Participate in a production support and on-call rotation, taking ownership of the most complex issue resolution and driving root-cause analysis that improves system resiliency. You Have 10+ years of experience building and operating large-scale distributed systems, data platforms, or observability infrastructure, including time at Principal or Staff level. Deep Observability Expertise: Hands-on production ownership of metrics, logs, and distributed tracing at scale, including at least one major backend (Datadog, or comparable) and a working understanding of cardinality and cost mechanics. Telemetry Data Engineering: Demonstrated depth in large-scale data architecture, including open table formats (Delta Lake, Apache Iceberg), Spark or comparable distributed processing, streaming ingestion, partitioning and schema evolution, and query performance and cost tuning over very large datasets. Databricks Depth: Practical experience with the Databricks platform (jobs, clusters, Unity Catalog, Delta) and with MLflow for model and agent telemetry. OpenTelemetry Depth: Practical experience with OTel collectors, semantic conventions, context propagation, and sampling strategy, including tail-based sampling. Pipeline Engineering: Experience with high-volume log and telemetry pipelines, including FluentBit or Fluentd, streaming transport such as Kinesis or Kafka, and search backend index and mapping design. Advanced AWS & Kubernetes Expertise: EKS, ECS Fargate, EC2, Lambda, and CloudWatch in production. 10+ years of programming experience with modern languages such as Go, Java, Python, or Scala, and strong SQL. Infrastructure as Code: Terraform, and GitOps workflows such as Flux or ArgoCD. Architectural Influence at Scale: A track record of setting technical direction that multiple teams and organizations adopted, including the written artifacts (architecture decisions, standards, RFCs) that made it durable after you moved on. Strong incident response instincts, with experience improving mean-time-to-resolution through better instrumentation and better data rather than more heroics. A degree in Computer Science, Engineering, or a related field, or equivalent practical industry experience. Legally eligible to work in the U.S. on an ongoing basis Nice to Have Experience instrumenting LLM or agentic systems, including OTel GenAI semantic conventions and tracing agent tool-call workflows. Data reliability engineering practice: freshness, quality, and lineage SLOs for production data pipelines Telemetry cost engineering or FinOps at scale. Experience in regulated environments (FedRAMP, GovCloud) and designing telemetry that remains useful under redaction. Prometheus, Grafana, and Alertmanager. Snowflake, or experience operating across more than one lakehouse or warehouse platform. Service catalog or internal developer platform work (Backstage or similar). Current US Perks & Benefits: Employer subsidized medical/v