Next steps by user role（按用户角色划分的后续步骤）¶
Now that you are oriented in the Palantir platform and understand its core concepts, you can explore the platform capabilities that are most relevant for your role.
The boundaries between roles in the Palantir platform can shift, and some responsibilities and workflows may not align perfectly into a single role. You may end up performing different roles; if your organization is just getting started with Palantir, or if you are working on a small implementation team, you may use many parts of the platform in your day-to-day work.
Alternatively, you can explore the Palantir platform by reviewing the application reference, which provides a high-level overview of the major platform applications.
Standard user roles¶
With this in mind, the following roles are generally adopted at most organizations using Palantir. Below, we discuss these high-level roles and how each type of user can get started:
If you would like to get started right away, jump into the Palantir platform.
Data engineer¶
Palantir's data integration layer provides the foundation for all the other work that happens in the platform. By building and maintaining data pipelines, data engineers produce datasets that are high-quality, relevant, and frequently updated to serve the needs of the organization. A wide variety of tools are available to maintain the durability of data pipelines over time, including programmatic health checks and transparency into the underlying computation.
The primary tools used by data engineers include Pipeline Builder and Code Repositories for authoring data pipelines, and Data Lineage for visualizing them end-to-end. Data engineers should also understand how to use recommended health checks, monitoring views, and lineage to operate pipelines after they are deployed. Data engineers will need to be familiar with the concept of data pipelines and develop an understanding of what makes for a high-quality pipeline in the platform.
Learn more about data pipelines.
Application builder¶
Palantir's Ontology and application building capabilities enable you to create tailored applications for end users. These end users are usually operators in an organization, making decisions that can be informed by data. Beyond just presenting data to users, you can use custom applications to capture information from users in the form of action types configured in the Ontology.
Application builders will need to be familiar with the Palantir Ontology, which is usually the layer at which application builders collaborate with data engineers to establish a data foundation for workflow development. Builders can create and maintain their organization's Ontology in Ontology Manager, then use Workflow Lineage to understand how objects, actions, functions, large language models (LLMs), and applications fit together.
To create and deliver applications, builders can use Workshop for point-and-click application building on top of the Ontology, extend Workshop with custom widgets, or use Pilot to start from a natural language description. For code-first development, builders can use the OSDK and OSDK React applications, manage application configuration and SDK generation in Developer Console, write Functions for shared business logic, or build applications in Slate. Builders can also create LLM-backed workflows with AIP Logic and trigger ontology-driven work with Automate.
Learn more about application building.
Data scientist¶
The Palantir platform includes support for analyzing data using code and developing, evaluating, and deploying machine learning models. This functionality builds on top of the rigor of the data integration layer to provide lineage and reproducibility for models in the same way as datasets. The result is an environment where analytics and machine learning can build on high-quality data and shared modeling workflows.
In the Palantir platform, data scientists often use Code Workbook, an application designed to enable code-based analysis and the development of machine learning models. Code Workbook enables you to write code in Python, R, and SQL to access, normalize, and analyze high-quality datasets prepared by data engineers. The resulting analyses and models can then be connected to model integration, shared through Model Catalog, evaluated with AIP Evals, and integrated into the Ontology for use in applications and workflows.
As an alternative, data scientists can work in their preferred third-party IDEs with Code Workspaces. Code Workspaces containers are integrated with the rest of the Palantir ecosystem to combine JupyterLab® and RStudio® Workbench IDEs with the security, branching, and resource management benefits of the Palantir platform.
Learn more about model integration and code-based analysis.
Analyst¶
As Palantir can be used to build a secure and high-quality data foundation, analysts can find and explore data that is relevant to the questions they need to answer. A rich set of tools is available for analyzing data in a wide variety of formats—tabular, relational, temporal, geospatial, and more. Once your analysis yields insight, you can make it repeatable by creating dashboards or present your findings using reporting tools.
Analysts typically use Contour to explore datasets in the platform and conduct open-ended analysis at high-scale, and use Quiver to analyze data in the Ontology, along with associated time series. Analysts can also use AIP Analyst to explore ontology-backed questions with natural language. These applications support moving from ad-hoc analysis into dashboards, reports, or Notepad documents that share results with colleagues.
Platform administrator¶
Platform administrators can use Palantir's dedicated administrative tooling to configure the platform, manage and understand how it is being used, and ensure that the organization's data is being managed securely.
Platform administrators typically set up authentication to connect to an organization's identity provider, configure application access, then set up Data Connection to enable data to flow into the platform. As use of the platform matures, administrators can use Resource Management to manage resource consumption, Developer Console to support custom application development, and monitoring views to help teams observe operational health. Administrators should also understand access considerations for Palantir MCP and Ontology MCP when those capabilities are enabled.
Learn more about platform administration.
Data governance¶
Palantir provides tools for securing data and providing transparency to data governance leads. These tools provide guarantees about how data is protected as it is transformed in the Palantir platform and used for user-facing applications. They also preserve the ability for you to introspect and validate who has access to what information.
Users in data governance roles should learn about the broad set of data security workflows in the platform, ranging from securing a data foundation to protecting sensitive data. These capabilities are built on Palantir's data security concepts, namely Projects and Markings. Governance leads should also understand object permissioning, application restrictions, OSDK permissions, Ontology MCP restrictions, and Workflow Lineage security visibility where those controls are part of their organization's workflows.
Learn more about data protection and governance.
Jump into the platform¶
Now that you have learned how to navigate through the platform, start building by:
- Learning with the Training application, a curated set of courses from learn.palantir.com.
- Using Examples, a comprehensive library designed to facilitate learning and effective platform usage with reference examples, tutorials, and starter kits.
中文翻译¶
按用户角色划分的后续步骤¶
现在您已熟悉Palantir平台并理解其核心概念，可以探索与您角色最相关的平台功能。
Palantir平台中各角色之间的界限可能有所变化，某些职责和工作流程可能无法完全对应到单一角色。您可能会承担不同角色的工作；如果您的组织刚刚开始使用Palantir，或者您在一个小型实施团队中工作，日常工作中可能会用到平台的多个部分。
此外，您也可以通过查阅应用参考文档来探索Palantir平台，该文档提供了主要平台应用的高层概览。
标准用户角色¶
基于以上说明，大多数使用Palantir的组织通常采用以下角色。下面我们将讨论这些高层级角色以及各类用户如何开始使用：
如果您希望立即开始使用，请直接进入Palantir平台。
数据工程师¶
Palantir的数据集成层为平台中所有其他工作提供了基础。通过构建和维护数据管道，数据工程师能够生成高质量、相关性强且频繁更新的数据集，以满足组织的需求。平台提供了多种工具来维护数据管道的持久性，包括程序化健康检查和对底层计算的透明性。
数据工程师使用的主要工具包括用于编写数据管道的Pipeline Builder和代码仓库(Code Repositories)，以及用于端到端可视化的数据沿袭(Data Lineage)。数据工程师还应了解如何使用推荐健康检查、监控视图(Monitoring Views)和沿袭功能来运维已部署的管道。数据工程师需要熟悉数据管道的概念，并理解如何在平台中构建高质量管道。
应用构建者¶
Palantir的本体论(Ontology)和应用构建能力使您能够为最终用户创建定制化应用。这些最终用户通常是组织中的操作人员，他们需要基于数据做出决策。除了向用户展示数据外，您还可以通过在本体论中配置的操作类型(Action Types)来捕获用户输入的信息。
应用构建者需要熟悉Palantir的本体论(Ontology)，这通常是应用构建者与数据工程师协作建立工作流开发数据基础的层级。构建者可以在Ontology Manager中创建和维护组织的本体论，然后使用工作流沿袭(Workflow Lineage)来理解对象、操作、函数、大语言模型(LLMs)和应用之间的关联。
为了创建和交付应用，构建者可以使用Workshop在本体论之上进行点击式应用构建，通过自定义组件(Custom Widgets)扩展Workshop，或使用Pilot从自然语言描述开始构建。对于代码优先的开发，构建者可以使用OSDK和OSDK React应用，在开发者控制台(Developer Console)中管理应用配置和SDK生成，编写函数(Functions)以实现共享业务逻辑，或在Slate中构建应用。构建者还可以使用AIP Logic创建基于LLM的工作流，并通过Automate触发本体论驱动的工作。
数据科学家¶
Palantir平台支持使用代码分析数据，以及开发、评估和部署机器学习模型。该功能建立在数据集成层的严谨性之上，为模型提供与数据集相同的沿袭和可复现性。最终形成一个环境，使分析和机器学习能够基于高质量数据和共享建模工作流进行。
在Palantir平台中，数据科学家通常使用Code Workbook，这是一个专为基于代码的分析和机器学习模型开发而设计的应用。Code Workbook允许您使用Python、R和SQL编写代码，以访问、标准化和分析由数据工程师准备的高质量数据集。生成的分析结果和模型可以连接到模型集成(Model Integration)，通过模型目录(Model Catalog)共享，使用AIP Evals进行评估，并集成到本体论中，供应用和工作流使用。
作为替代方案，数据科学家可以在他们偏好的第三方IDE中使用Code Workspaces进行工作。Code Workspaces容器与Palantir生态系统的其他部分集成，将JupyterLab®和RStudio® Workbench IDE与Palantir平台的安全性、分支管理和资源管理优势相结合。
分析师¶
由于Palantir可用于构建安全且高质量的数据基础，分析师可以查找和探索与需要解答的问题相关的数据。平台提供了丰富的工具集，用于分析各种格式的数据——表格型、关系型、时间序列、地理空间等。一旦分析得出洞察，您可以通过创建仪表盘(Dashboards)使其可复现，或使用报告(Reporting)工具展示您的发现。
分析师通常使用Contour在平台中探索数据集并进行大规模开放式分析，使用Quiver分析本体论中的数据及其相关时间序列。分析师还可以使用AIP Analyst通过自然语言探索基于本体论的问题。这些应用支持将临时分析转化为仪表盘、报告或Notepad文档，与同事分享结果。
平台管理员¶
平台管理员可以使用Palantir的专用管理工具来配置平台、管理和了解平台的使用情况，并确保组织的数据得到安全管理。
平台管理员通常设置身份验证(Authentication)以连接到组织的身份提供商，配置应用访问权限，然后设置Data Connection使数据能够流入平台。随着平台使用的成熟，管理员可以使用资源管理(Resource Management)管理资源消耗，使用开发者控制台(Developer Console)支持自定义应用开发，以及使用监控视图(Monitoring Views)帮助团队观察运营健康状况。当启用Palantir MCP和Ontology MCP功能时，管理员还应了解相关的访问注意事项。
数据治理¶
Palantir提供了保护数据并为数据治理负责人提供透明度的工具。这些工具确保数据在Palantir平台中转换以及用于面向用户的应用时得到保护。它们还保留了您检查和验证谁有权访问哪些信息的能力。
数据治理角色的用户应了解平台中广泛的数据安全工作流，从保护数据基础到保护敏感数据。这些能力建立在Palantir的数据安全概念之上，即项目(Projects)和标记(Markings)。治理负责人还应了解对象权限管理(Object Permissioning)、应用限制(Application Restrictions)、OSDK权限、Ontology MCP限制以及工作流沿袭安全可见性，当这些控制措施是其组织工作流的一部分时。
直接进入平台¶
现在您已了解如何在平台中导航，可以通过以下方式开始构建：
- 通过培训应用(Training Application)，来自learn.palantir.com的精选课程集进行学习。
- 使用示例(Examples)，这是一个全面的库，旨在通过参考示例、教程和入门套件促进学习和有效使用平台。