To get the best candidate experience, please consider applying for a maximum of 3 roles within 12 months to ensure you are not duplicating efforts.

Job Category

Job Details

The Role
We're hiring an AI Cloud Architect to help shape how agentic systems are built, deployed, and scaled at Salesforce. You'll work at the frontier of the Agentforce platform — inventing new primitives, incubating next-gen agentic systems, partnering directly with customers, and setting architectural direction for how enterprise AI agents work in production.

This is a builder-communicator role. You'll write code and whitepapers, design runtimes and present to executive leadership.

Invent and Incubate
Drive fundamental advances in agentic AI — new execution models, agent orchestration frameworks, and stateful runtime architectures. Lead incubation projects exploring next-generation agentic systems for the enterprise, including OpenClaw-style reasoning architectures and emerging multi-agent coordination patterns. The kind of work that produced Agent Graph, AgentScript, and the Graph Runtime powering Agentforce today.

Shape the Core Platform
Influence core architectural decisions across Agentforce — agent determinism, A2A protocols, long-running async workflows, and context engineering. Contribute adjacently to the agent development lifecycle — including Agent Builder, Agent Testing Center, and Agent Observability — helping define how developers build, validate, and monitor agents end-to-end.

Partner with customers
Work directly with strategic customers to understand their business context and co-design solutions using Agentforce. Translate complex agent capabilities into concrete, trustworthy outcomes — from proof-of-concept to production.

Write and communicate with conviction
Produce technically rigorous, persuasive content — design docs, engineering blog posts, developer documentation, and external presentations. You have a clear point of view and can defend it.

• Expert in distributed systems, stateful runtimes, and LLM-based agent architectures

• Experience building orchestration frameworks, graph-based execution engines, or agent DSLs

• Strong grasp of LLM behavior, context management, and agent reliability in production

• Fluency in Python and Java; comfortable with cloud-native infra

• Curiosity about and familiarity with frontier agentic research — reasoning loops, tool use, memory architectures, and multi-agent systems

• You've shipped durable technical artifacts — platforms, frameworks, or open-source projects others build on

• Comfortable going from early-stage experiment to prototype to GA in fast-moving environments

• Experience with incubation, 01 projects, or founding engineer environments is a strong plus

• Exceptional written communication — from design docs to customer whitepapers

• Can speak to a developer, an enterprise architect, and a CIO with equal clarity and credibility

• Experience engaging enterprise customers as a trusted technical advisor

• Drives alignment across engineering, product, and field without requiring authority

• Raises the technical bar around them — a natural teacher and force multiplier

• Open-source community experience or developer advocacy background

• Familiarity with Salesforce platform primitives (Flow, Apex, Einstein, metadata)

• Background in ML infrastructure, data engineering, or AI platform work

• Experience with language/compiler design (relevant to DSL authoring systems like AgentScript)

• Familiarity with observability tooling, testing frameworks, or developer experience platforms

Why This Role
The agentic AI space is moving fast, and the hardest problems — agent determinism, stateful long-running workflows, multi-agent coordination, enterprise-grade reliability, next-gen reasoning architectures — don't have solved answers yet. You'll work alongside exceptional systems thinkers and lead incubation work that could define the next chapter of Agentforce, while also contributing to the platform that already ships to millions of customers. It's rare to find a role that demands both deep invention and real-world enterprise impact. This one does.