Software Engineering Architect, Salesforce

• Experience: 8+ years of industry experience, with a proven track record of leading large-scale distributed systems or platform engineering
• AI/ML Proficiency: Deep understanding of the LLM lifecycle—including prompt engineering, RAG (Retrieval-Augmented Generation), and agentic reasoning patterns (e.g., ReAct, Chain-of-Thought)
• Languages: Strong proficiency in Java, Go, or Python. Familiarity with Hack/PHP (Slack’s core) or TypeScript is a major plus
• Cloud & Infrastructure: Experience with Kubernetes, Docker, and managing integrations with major AI providers (OpenAI, Anthropic, or Salesforce’s own models via Einstein)
• Data Systems: Familiarity with vector databases (Pinecone, Milvus), event-driven architecture (Kafka), and large-scale data processing (Data Cloud)

Desirable

• As an Architect of Agent Systems in the Slack Agent Systems group, you will be a primary architect and technical lead for the next generation of work automation. You will be building the system that provides framework, guardrails, and overall management to enable agents and humans to work seamlessly together in slack
• The Slack Agent Systems team sits at the most exciting intersection in tech today: the point where Generative AI meets the place where work happens. You will have the unique opportunity to define how millions of people interact with AI agents every single day
• You will lead the development of the Agent SDK and orchestration layers that allow agents to live natively within Slack, transforming the platform into the agentic OS
• Architect Agentic Orchestration: Design and implement high-scale systems that manage agent reasoning (LLM loops), planning, and tool-calling (MuleSoft, Data Cloud, and third-party APIs)
• Lead Technical Strategy: Define the roadmap for "Human-in-the-loop" (HITL) checkpoints, ensuring agents operate safely within enterprise boundaries while maintaining high autonomy
• Build the SDK: Own and evolve the developer-facing frameworks that allow internal and external teams to build, test, and deploy Slack Agents at scale
• Optimize Performance: Solve for low-latency inference at "Slack scale," managing GPU-accelerated workloads and real-time event streaming across millions of active channels
• Cross-Functional Leadership: Partner with Product, Security, and Data Science to move Slack AI from "summarization" to "action," ensuring agents have the correct long-term memory and shared state
• Mentor & Standardize: Set the "Gold Standard" for engineering craft, conducting deep design reviews and advocating for AI-first development patterns (e.g., automated evaluation frameworks for LLM outputs)