Agentforce Architect

Role overview Establish the patterns and governance for a new agentic architecture discipline inside a North American client's stack, deciding which processes warrant an agent versus deterministic automation versus neither, and how agents are bounded, grounded, handed off, and audited. This remote, full-time contractor contributor role is delivered from Latin America, working the client's core business hours with substantial time-zone overlap, partnering with the client's squad under delivery leadership.

Responsibilities - Decide which business processes should be agentified, which warrant deterministic automation, and which should be declined, including telling executives when a requested use case will not work. - Design agent topology, including agent boundaries, topic decomposition, action inventories, escalation paths, and the grounding architecture across data, knowledge, and retrieval sources. - Define the trust and governance standard covering action authorization, PII handling, injection defense, audit trail, and the human-in-the-loop criteria that trigger handoff. - Establish evaluation and release gates that agents must clear before facing customers, including red-teaming and regression evaluation. - Guide and review the work of AI and data engineers, set the measurement model for containment, resolution quality, and cost per interaction, and advise executives on realistic scope, cost, and deflection expectations. - Design how agents sit alongside existing Service Cloud, Sales Cloud, and integration architecture, and own the consumption-cost architecture for the agent platform and its data-layer dependencies.

Requirements - Bachelor's degree in Computer Science, Software Engineering, or a related field, or equivalent practical experience. - 8+ years in solution or technical architecture, including hands-on delivery of production LLM or agentic systems from concept through measured outcomes. - Deep expertise with agent-builder tooling, topic and action design, prompt building, and trust-layer and reasoning components in agentic platforms. - Strong command of a major data and retrieval platform used as a grounding layer, plus Service Cloud and Sales Cloud process design. - Practical AI safety chops: guardrails, red-teaming, regression evaluation, and observability practice. - Professional English with the ability to run requirement and design conversations directly with North American stakeholders.

Nice to have - Salesforce Certified AI Specialist, plus Application Architect or System Architect credentials.

Benefits and work setup - Remote full-time contractor arrangement, based in Latin America, working North American core hours with substantial overlap; visa sponsorship is not provided.