Start with the work: generate invoices, enrich leads, test an application or build a model. People across functions can configure and build AI-enabled workflows, with engineering support matched to the task. Define what changes, the work product and who checks it. The role map below helps connect those responsibilities to your enterprise.
44 role groups · 264,613 real job postings · See roles employers are hiring for →
Work transformation in practice
Build on the work your people already do. These examples show how software delivery, testing and business teams could add AI capabilities, what they would produce and what needs checking.
Software engineering and QA
IT services and software delivery
Finance operations
Illustrative examples for scoping, not customer results or a fixed catalogue.
Choose a technical project around the work your enterprise needs to do. These illustrative starting points require different tools, data and engineering depth.
Define the user journeys and expected behavior first. Give the agent an approved test environment, accounts and fixtures. Build the test flow, then review assertions, reproducible defect reports, missed defects and false alarms. Keep release acceptance with the responsible team.
Read the practical guide →Discuss this work →Start with an authorized target and agreed testing boundaries. Scope the access, permitted checks and stop conditions. Review findings against known cases, verify evidence and prioritize remediation with the security owner. A practice result does not establish that a site is secure.
Discuss this work →First decide whether the problem needs model training, retrieval, an existing model or a deterministic system. A model-building project needs an agreed task, usable data, a baseline, compute and evaluation on held-out examples. Training from scratch is a different scope from adapting an existing model.
Discuss this work →Define the behavior to improve and prepare suitable examples with appropriate data rights. Separate training and evaluation data, compare the adapted model with the baseline, and inspect failure cases, cost and operating constraints. Fine-tuning alone does not keep business facts current or establish correctness.
Read the practical guide →Discuss this work →Business expertise shapes what gets built: invoice rules in Finance, lead qualification in Sales, contract review in Legal. These role patterns help identify responsibilities. The enterprise project determines the tools, technical support and evidence needed.
Builds or configures AI workflows for campaign planning, segmentation, content operations and creative production.
Builds or configures AI assistance for financial planning, accounting, audit and treasury work.
Builds or configures AI workflows for recruitment, learning, performance and internal mobility.
Builds or configures AI assistance for contract review, compliance monitoring and regulatory writing.
Builds or configures AI workflows for lead research, outreach preparation and sales follow-up.
Builds or configures AI workflows for operations, supply chain, logistics, procurement and fulfilment.
Builds or configures AI assistance for customer service, success and support.
Maps business requirements, records and decisions for an AI project.
Designs interfaces and interactions through which people use AI systems.
Helps customers use and review AI in their work after implementation.
Produces, edits and reviews content with AI tools.
Technical work includes model development, fine-tuning, agent construction, application integration, testing and delivery into customer systems. Match the engineering depth to the problem and the enterprise constraints; start from the capabilities people already bring.
Builds applications and features using AI models, tools and enterprise data.
Applies AI models to a customer or business problem.
Works within the customer's workflow to connect a business problem to a working technical solution.
Embeds in a business function to build the agents and workflows that team uses.
Builds workflows that connect agents, tools and multiple steps.
Builds agents that use tools, manage context and act within defined permissions.
Designs, tests, and maintains prompt suites and evals.
Builds the loops, tools and evaluation harnesses that agents run inside.
Trains, evaluates and deploys machine-learning models.
Builds, adapts and operates applications and services around large language models.
Designs the architecture and integrations for an AI system, including model services, data, permissions, observability and failure handling.
Develops and tests ideas for model architectures, training methods and evaluation.
Builds the data layer AI runs on: ingestion, feature stores, embeddings, retrieval.
Connects an AI implementation to the customer’s operating systems.
Builds and maintains model-serving, agent-runtime and evaluation infrastructure used by delivery teams.
Reviews AI systems for prompt injection, model supply-chain risks, agent permissions and data boundaries.
Leads the operation of machine-learning systems, including deployment pipelines, monitoring, retraining and rollback.
Helps customers evaluate and implement an AI product.
Some roles coordinate work across teams. AI leaders set priorities and assign responsibility. Responsible AI roles define governance and review risks. AI evangelists explain capabilities to users and partners. The role map also includes people building software with AI without a conventional engineering background.
Enterprise AI leader responsible for the organization's AI priorities and governance.
Leads AI work for a function or organization.
Enterprise AI leadership at the VP/SVP level, typically running the AI platform, AI product, or AI-X function (AI Security, AI Strategy, AI Innovation).
Operational AI ownership, often the role accountable for shipping AI features and AI-touching products in a specific business unit.
Sets AI governance policy and coordinates risk review across the organization.
Explains AI capabilities through demonstrations, technical content and workshops.
Uses natural-language instructions and AI coding tools to build software.
Coordinates agents, tools and human decisions across a workflow.
Checks AI outputs and agent actions against defined criteria.
Defines what an AI product should help users do and which decisions remain with its owners.
Develops AI priorities with business and technical owners.
Coordinates the delivery and introduction of AI initiatives across teams.
Helps an organization define an AI problem, compare approaches and plan the work.
Helps teams introduce AI into agreed responsibilities.
Monitors agent queues, outputs, failures and escalations after implementation.
Some AI work is done by contracted domain experts. They review model answers, label examples, design tests and investigate failures. These roles sit outside the employee role map and can require substantial subject expertise.
Bring a real workflow and the people who need to change it. Scope a simulation around your systems, data and rules, with agreed building responsibilities and checks. The AI Bootcamp is one 6 to 8 week preparation path; established labs and direct agent evaluation remain available.
