Software Engineering Architect

Enboarder; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Salesforce is seeking an Architect to lead the strategy and technical direction for its Internal Agentic Platform, MCP, and Model Gateway. You will shape how agents connect to tools, models, and enterprise knowledge, enabling secure, reusable integrations across the SDLC and production environments.

You will drive adoption of Mastra AI and LangChain/LangGraph, ensuring coherence, extensibility, and cost-effective operation while guiding teams from prototypes to dependable production agents.

Qualifications

• 15+ years designing and delivering large-scale distributed systems
• Track record of setting strategy and directing implementation
• Hands-on leadership in agentic platform architecture
• Experience with Mastra AI and LangChain/LangGraph
• Experience designing model gateways and inference platforms
• Understanding QoS: latency, throughput, reliability, cost tradeoffs
• Familiar with SLMs/LLMs, model routing, quantization, and optimization
• Experience with enterprise knowledge or retrieval systems (RAG/Graph RAG)
• APIs and extensible integration patterns
• Judgment for security, privacy, reliability, and performance

Responsibilities

• Own strategy and architecture for a powerful agentic platform experience across Salesforce
• Define reusable platform capabilities for agent orchestration and tool integration
• Set technical direction for agentic framework adoption (Mastra AI, LangChain/LangGraph)
• Establish patterns for production-grade agents with observability and evaluation
• Drive AI-enabled developer experiences across the SDLC (code understanding, generation, review)
• Define how MCP fits into runtimes and integrations with enterprise tools
• Architect secure, extensible LLM Model Gateway with governance across providers
• Guide model portfolio decisions across hosted, open-weight, fine-tuned models
• Develop metrics for model economics, cost per task, and throughput
• Align platform strategy with engineering, product, security, and data teams
• Prototype ideas and validate architectures through working systems
• Communicate decisions clearly and mentor engineers

Skills

Strategic architecture

Agentic platform

MCP | Model Context Protocol

Mastra AI

LangChain / LangGraph

APIs & integration patterns

Security & governance

Leadership & mentoring

SDLC & developer experience

Cost modeling & optimization

Education

Related technical degree

Tools

Mastra AI

LangChain

LangGraph

Job description

The Experience

Salesforce is transforming the Developer Experience using AI - to change how software is designed, built, tested, secured, and operated. We're looking for an Architect to set the strategy and technical direction for our Internal Agentic Platform, Model Context Protocol (MCP), LLM Model Gateway, and Context Engineering capabilities.

This is a hands-on architecture role for a technical leader who can define a compelling platform experience and help deliver it in production. You'll shape how agents work, how they connect to tools and models, how they use enterprise knowledge, and how they support developers across the software development lifecycle (SDLC). Your work will help teams adopt AI safely and effectively, improve developer productivity, and avoid solving the same platform problems independently.

What You'll Actually Be Doing

• Own the strategy and architecture for a powerful agentic platform experience that enables teams to build, compose, deploy, and operate agents across Salesforce. Set architecture and adoption direction for MCP, including server and client patterns, tool discovery, interoperability, identity, authorization, and lifecycle management.
• Define reusable platform capabilities for agent orchestration, tool and skill integration, agent runtimes, state and memory, human oversight, evaluation, and production operations.
• Set technical direction for agentic framework adoption, including the use of Mastra AI and LangChain/LangGraph. Guide framework integration and evolution while keeping the platform coherent, extensible, and adaptable to new approaches.
• Establish patterns that help teams move from prototypes to dependable production agents, with clear paths for customization, evaluation, observability, and ongoing improvement.
• Drive new AI-enabled developer experiences across the SDLC, including code understanding and generation, test creation, code review, debugging, modernization, documentation, and release or operational workflows.
• Define how MCP fits into agent runtimes and the broader platform, enabling agents to connect to enterprise tools and systems through secure, governed, and reusable integrations.
• Define the architecture and evolution of a secure, extensible LLM Model Gateway that provides consistent model access, policy enforcement, observability, resilience, and governance across providers and deployment types.
• Drive a cost-effective model portfolio spanning small language models (SLMs) and large language models (LLMs). Evaluate when to use hosted, open-weight, fine-tuned, or specialized models for different workloads.
• Develop ways to measure and improve model economics, including cost per successful task, token and context efficiency, retry rates, inference utilization, and the impact of model or routing changes.
• Own the strategy for context engineering and enterprise knowledge management so agents can access relevant, grounded, and permission-aware information.
• Guide the architecture of enterprise retrieval systems, including retrieval-augmented generation (RAG) and Graph RAG, and determine how they work with enterprise data, knowledge sources, and agent workflows.
• Set shared standards for security, privacy, governance, reliability, observability, evaluation, and human oversight across agents, models, tools, and data.
• Stay close to implementation by prototyping ideas, validating architecture through working systems, reviewing critical designs, and helping teams resolve technical tradeoffs.
• Align engineering, product, security, data, and infrastructure partners around platform strategy, adoption, and delivery.
• Communicate architectural decisions clearly, influence across teams, and mentor engineers and technical leaders.
• Design complex systems where AI agents fit smoothly into human workflows, and ship secure, high-quality software with AI as a core part of your own development workflow. Build and maintain the shared system context that lets AI work accurately, and review code, whether written by people or AI, for correctness, quality, security, and performance.

You're Our Person If...

• You have 15+ years of experience designing and delivering large-scale distributed systems, developer platforms, AI platforms, or infrastructure services.
• You have a track record of setting strategy and technical direction across teams, then carrying it through implementation and production operation.
• You have strong expertise in agentic platform architecture and experience creating platform capabilities that make agent development and operation effective for application teams.
• You have hands-on experience with agentic frameworks such as Mastra AI and LangChain/LangGraph, including the ability to evaluate their tradeoffs and establish scalable platform patterns.
• You have experience designing model gateways, inference platforms, model-serving systems, model orchestration layers, or equivalent shared services.
• You have experience evaluating models against real workloads, with a practical understanding of quality, latency, throughput, reliability, and cost tradeoffs.
• You are familiar with SLMs, LLMs, model routing, fine-tuning, quantization, inference optimization, or other ways to improve cost and performance.
• You have experience building enterprise knowledge or retrieval systems using RAG, Graph RAG, search, or related approaches.
• You have experience designing application programming interfaces (APIs) and extensible integration patterns. Familiarity with MCP or comparable tool and protocol ecosystems is a plus.
• You have expertise in one or more relevant areas, such as identity and authorization, data platforms, retrieval and search, context management, workflow orchestration, platform security, or developer experience.
• You understand software development workflows and how to assess whether AI capabilities improve developer experience, productivity, and engineering outcomes.
• You use sound judgment around security, privacy, reliability, performance, and operational cost in fast-changing technical environments.
• You have strong communication skills and the ability to build alignment across engineering, product, security, and business teams.
• You bring a genuine AI-first approach to engineering. You use AI tools such as Claude Code, GitHub Copilot, Codex, or Cursor to move faster and work across the stack, and you write precise, structured prompts that make AI output reliable, secure, and production-ready.
• A related technical degree required.

Even Better If...

• You have experience delivering production LLM or agent systems, including evaluation, telemetry, safety, and operational feedback loops.
• You have experience with coding assistants, AI-enabled developer tools, continuous integration and delivery (CI/CD) workflows, automated testing, code review, or software engineering productivity initiatives.
• You have experience building internal platforms, paved paths, or developer tools adopted by multiple teams.
• You have experience with multi-model architectures spanning commercial APIs, open-weight models, and self-hosted or specialized models.
• You have contributed to technical standards, open-source projects, or cross-organization architecture initiatives.