Principal Architect

Salesforce Salesforce · Enterprise · San Francisco Metro -, Ohio -, California - Palo Alto, North Carolina -, Illinois - Chicago, Washington D.C. -, California -, Washington - Bellevue, California - Irvine, Indiana - Indianapolis, California - San Francisco, Arizona -, New York - New York, Utah -, Minnesota -, Washington - Seattle, Texas - Dallas, Washington -, Georgia - Atlanta, US, CA +1 · Remote

Salesforce is seeking a Principal Architect to lead their AI-native engineering organization, focusing on architecting and scaling systems where human intellect and autonomous agents collaborate. The role involves defining the AI lifecycle, building agent infrastructure, establishing rigorous evaluation frameworks, and pioneering technical taste for agent-accelerated development. Requires deep experience in large-scale systems and cutting-edge AI architectures, particularly agentic loops and multi-agent orchestration.

What you'd actually do

  1. Define, build, and scale the foundational framework, review standards, and compilation/validation loops for autonomous, agentic software development across the entire enterprise.
  2. Turn highly ambiguous, long-term business goals into concrete, mathematically sound architectural plans, spec-driven development methodologies, and deterministic validation strategies.
  3. Design advanced system topologies, multi-agent orchestration frameworks, and context-bounding mechanisms that allow AI agents to safely operate on large, complex codebases.
  4. Lead the creation of next-generation evaluation harnesses, golden test suites, semantic CI/CD gates, and real-time behavioral monitoring to guarantee the safety, security, and deterministic correctness of agentic output.
  5. Act as the ultimate arbiter of code quality, structural integrity, and architectural purity, ensuring that agent-accelerated development reduces technical debt rather than compounding it.

Skills

Required

  • Systems architecture
  • LLM capabilities
  • cognitive architectures
  • enterprise-grade software engineering
  • agentic loops
  • reinforcement learning
  • cloud platforms
  • advanced AI architectures
  • large-scale distributed systems
  • enterprise architectures
  • foundational infrastructure
  • fundamental computer science
  • memory management
  • distributed state
  • concurrency models
  • Rust
  • Python
  • Go
  • C++
  • TypeScript
  • mission-critical production systems
  • high availability
  • strict security postures
  • predictable performance characteristics
  • multi-agent orchestration
  • context-bounding mechanisms
  • agentic identity
  • trust
  • compliance
  • governance
  • evaluation harnesses
  • golden test suites
  • semantic CI/CD gates
  • real-time behavioral monitoring
  • safety
  • security
  • deterministic correctness
  • code quality
  • structural integrity
  • architectural purity
  • technical debt reduction
  • Executive Leadership partnership
  • Product partnership
  • Security partnership
  • Core Infrastructure partnership
  • thought leadership
  • mentoring Principal and Staff engineers
  • rigorous engineering judgment
  • ethics for AI use
  • industry standards
  • open source code
  • human-AI interaction
  • creative AI applications
  • prompt engineering
  • IDE extensions
  • agentic workflows
  • autonomous code-generation loops
  • LLM architectures
  • context-window dynamics
  • RAG systems
  • function-calling mechanics
  • fine-tuning paradigms
  • software engineering tasks

Nice to have

  • Ph.D. in Computer Science/AI
  • M.S. or Ph.D. in Artificial Intelligence, Machine Learning, or a deeply quantitative field
  • symbolic engineering
  • probabilistic AI
  • hyper-dense experience
  • advanced research degree
  • customized production-grade agentic workflows
  • multi-agent choreographies
  • cognitive architecture fluency
  • frontier LLM architectures
  • open source LLM architectures
  • context-window dynamics
  • RAG systems
  • function-calling mechanics
  • fine-tuning paradigms
  • software engineering tasks
  • deterministic systems for non-deterministic models

What the JD emphasized

  • 15 to 20+ years of building massive production systems
  • veteran engineering judgment
  • cutting-edge expertise in agentic loops
  • architectural breakthroughs
  • mathematically sound architectural plans
  • deterministic validation strategies
  • multi-agent orchestration frameworks
  • safely operate on large, complex codebases
  • rigorous evals
  • guarantee the safety, security, and deterministic correctness
  • Flawless Engineering Judgment
  • architecting mission-critical production systems
  • high availability
  • strict security postures
  • predictable performance characteristics
  • Production-Grade Agentic Experience
  • multi-agent choreographies
  • autonomous code-generation loops
  • Deterministic Systems for Non-Deterministic Models

Other signals

  • AI-native engineering organization
  • agentic loops
  • autonomous agents
  • multi-agent orchestration