Required Skills & Experience
- Agentforce Builder: agent, topic, instruction, and action design
- Prompt Builder templates grounded in record data
- Einstein Trust Layer data masking and audit trail
- Apex invocable actions and Flows as agent actions
- Data model, SOQL, and the sharing / permission model that limits what an agent can access
- External systems as actions via Named Credentials and External Services
- Deployment to Messaging for In-App and Web or Experience Cloud
- Source control with Git (branching, merge / rebase, pull requests)
- Testing Center suites for topic and action selection, rerun after every change
- Salesforce CLI deployment of agents, prompts, and actions
- Agent guardrails: what it must decline, when it hands off to a person, and how both are tested
- Secure handling of regulated government data, including what may reach an LLM prompt
- Salesforce Certified Agentforce Specialist
Nice to Have Skills & Experience
- Experience delivering Agentforce in Government Cloud Plus orgs
- Data 360 (Data Cloud) search indexes and retrievers for RAG grounding
- Agentforce DX and Agent Script for source-driven agent development
- Invoking MuleSoft-managed APIs as agent actions
- LWC for custom action inputs and response rendering
- Agent API for headless agent invocation
- Azure DevOps pipelines and Artifacts for CI/CD
- Automated, multi environment deployment configuration
- Post-launch tuning from session traces and Agentforce analytics
- Government Cloud constraints, including Agentforce feature gaps vs. commercial orgs
- CJIS Security Policy familiarity
- State and local government process context (licensing, case management)
- Salesforce Certified Platform Developer
- Salesforce Certified Platform Administrator
- Salesforce Certified Data Cloud Consultant
- Salesforce Certified Agentforce Service Consultant
- Salesforce Accredited Agentforce 360 for Public Sector Professional
Job Description
This role is split roughly evenly between new agent development and maintaining a high volume production agent. On the build side, the consultant designs, builds, and deploys Agentforce agents, typically in Government Cloud orgs: defining topics and instructions, building actions (Flow, Apex, and prompt templates), grounding responses in DPS data, and configuring the permissions that limit what each agent can access. Data held outside Salesforce is exposed as an action, typically through existing MuleSoft APIs or Named Credentials, and validated end to end. On the maintenance side, the consultant owns the health of the production agent: monitoring volume, handoff rates, action failures, and consumption against budget; reviewing session traces for misroutes and poor answers; and tuning topics, instructions, and grounding in response. At high volume a small instruction change can shift a large share of conversations, and Salesforce releases and model updates can change agent behavior on their own. Every change is regression tested before release, each Salesforce release is tested in a preview sandbox, and the consultant is the first point of triage when the agent misbehaves. Across both halves, the consultant is accountable for the things a larger team would normally distribute: gathering requirements directly with client stakeholders, setting realistic expectations about what an agent can reliably do, reporting agent performance to DPS on a regular cadence, managing the deployment pipeline and source control, and producing the design documentation and production support runbook the client will rely on after the engagement closes.