Principal Member of Technical Staff (PMTS) / Staff Software Engineer – AI Platform (Agentic Systems)

Salesforce•Atlanta, GA

2d•$197,300 - $344,700

About The Position

The Principal Member of Technical Staff / Staff Software Engineer will work on Slack’s Agentic Operations team to set technical direction and architecture for building, registering, and safely operating autonomous agentic systems. The role leads cross-team initiatives to transform high-volume software engineering workflows—including multi-repository code generation and automated incident remediation—into dependable, supervised, continuously improving agentic pipelines at enterprise scale. The Agentic Operations team builds and operates platforms and internal software factories that automate how Slack’s engineering organization develops, operates, and maintains services. Its systems support autonomous code generation and remediation; operational and incident automation; continuous evaluation, guardrails, and safety; durable execution and state management; and ongoing service-catalog and entity-metadata hygiene. The platform builds on Slack’s agentic development platform (Speck), its compute platform (EKS and EC2), and the service catalog. The team establishes patterns, guardrails, and operational standards that engineering teams across Slack can use to operate autonomous agents safely. The role also drives cross-functional adoption, cost-conscious model and compute utilization, clear ownership and accountability for failures, and reusable technical practices across the organization. It mentors senior engineers, leads architectural reviews, and authors system designs.

Requirements

10+ years of software engineering experience designing and delivering complex distributed systems at scale, with Staff/Principal-level technical judgment and organizational influence.
Hands-on experience building and shipping production LLM-powered agents and multi-agent systems, including orchestration, structured tool/function calling, retrieval-augmented generation (RAG), and context-window engineering.
Experience making non-deterministic AI systems production-ready through deterministic evaluation suites, regression harnesses, guardrails, and LLM telemetry.
Strong backend and distributed-systems engineering foundations, including clean, scalable, maintainable code and direct accountability for uptime, latency, and reliability.
A pragmatic approach to developer platforms and automation, comfort operating with ambiguity, and ability to balance AI capability with production safety.
Experience building autonomous code-generation pipelines, automated refactoring systems, or software factories that open pull requests and land changes across monolithic or multi-repository architectures.
Experience designing agent evaluation frameworks, prompt/context management layers, or LLM observability platforms used by other engineering teams.

Nice To Haves

Deep background in service orchestration or automated incident remediation at scale.
Hands-on experience with AWS, Kubernetes (EKS), and Infrastructure as Code (Terraform), using reusable, maintainable abstractions.
Active contributions to open-source agentic frameworks, developer tools, or the broader AI developer ecosystem.
Familiarity with frontier model APIs and agentic frameworks such as Anthropic Claude Tool Use, OpenAI function calling, Model Context Protocol (MCP), and LangGraph, and practical understanding of their failure modes.

Responsibilities

Set technical strategy and architect Slack’s agentic operations platform, including standards for safely building, registering, and operating autonomous agents and software factories at scale.
Design end-to-end autonomous pipelines for code generation, bug fixing, and incident remediation, with orchestration, tool integrations, and human-in-the-loop controls.
Build evaluation suites, regression harnesses, guardrails, and observability frameworks to make non-deterministic AI systems measurable, debuggable, and safe in production.
Define ownership and accountability models for detecting, attributing, and remediating autonomous-system failures, including on-call policies and operational guardrails.
Partner with platform teams, including Speck, Incident Management, and Deploy Safety, and consuming engineering groups to integrate agentic workflows into developer routines.
Promote cost-conscious and efficient model and compute utilization, balancing model intelligence and latency against token budgets and compute spend.
Mentor senior engineers, lead architectural reviews, and author system designs that establish reusable technical patterns across the organization.