Salesforce Forward Deployed Engineer
Confirmed on the employer's own hiring board on Oct 24, 2026. First seen by Alion on Oct 20, 2026.
Salesforce Forward Deployed Engineer
Build agents on Agentforce, extend them with custom MCP servers, and ship both alongside the people who use them.
EXPERIENCE
6-9 years in Salesforce
What you’ll do
A Forward Deployed Engineer works where the business problem actually lives. You’ll embed with a business team, learn their process in enough detail to argue about edge cases, and then build the thing - in their org, against their data, with them watching. The feedback loop is days, not quarters.
The work is a mix of three things. First, Agentforce: designing agents, topics, and actions on the platform, grounding them in Data Cloud, and tuning them until they behave in production. Second, custom MCP servers you design and write in Python, which connect agents (Agentforce and others) to systems and capabilities the platform doesn’t cover out of the box. Third, the core Salesforce engineering that makes both dependable: Apex, LWC, Flow, integrations, and a data model that holds up under real volume.
RESPONSIBILITIES
Run discovery directly with business stakeholders - process walkthroughs, org audits, data profiling - and turn what you learn into a design, not a backlog of tickets.
Design, build, and deploy Agentforce agents: topics, instructions, guardrails, and actions backed by Apex, Flow, prompt templates, and APIs.
Ground agents in the right data using Data Cloud, retrieval, and the Einstein Trust Layer, with an eye on accuracy, latency, and cost.
Decide, case by case, whether a capability belongs in a native Agentforce action or a custom MCP server, and build the connection between them when it’s the latter.
Design and build custom MCP servers in Python that expose Salesforce objects, SOQL queries, Apex endpoints, and business actions as well-scoped agent tools, with clear input schemas, useful error messages, and permissions that respect the org’s sharing model.
Build on-platform: Apex (triggers, Queueable, Batch, async patterns), Lightning Web Components, Flow, Platform Events, and a data model you’d defend in review.
Integrate Salesforce with external systems - REST and GraphQL APIs, OAuth 2.0 flows, Named Credentials, External Services, middleware - handling retries, idempotency, and rate limits properly.
Evaluate agent behavior end to end: write test cases, measure topic routing and action accuracy, and find failure modes before a user does.
Own deployment and quality - SFDX, scratch orgs, CI/CD, Apex test coverage that means something, and governor-limit awareness as a design habit rather than a bug fix.
Leave documentation and runbooks behind, so the team you embedded with can operate what you built after you move on.
REQUIREMENTS
Salesforce depth
6-9 years building on the Salesforce platform.
Production Apex at scale: bulkification, asynchronous patterns, governor limits, trigger frameworks.
Lightning Web Components and solid front-end fundamentals (JavaScript, HTML, CSS).
Declarative fluency - Flow, validation rules, sharing and visibility - and judgment about when not to write code.
Data modeling, large-data-volume considerations, and integration architecture.
Sales Cloud, Service Cloud, or Experience Cloud in production.
Agentforce, MCP, and Python
Hands-on Agentforce experience: building agents with topics and actions, Prompt Builder, Agent Builder, and Data Cloud grounding.
Strong Python - you’ve shipped and maintained services, not just scripts.
Hands-on experience building custom MCP servers or comparable LLM tool-use integrations (function calling, agent frameworks, retrieval pipelines).
A working mental model of how LLMs fail, and the habit of designing tools that fail safely.
API design, authentication, and secrets handling in an environment with real compliance constraints.
Comfort with Git, CI/CD, containers, and cloud deployment.
How you work
You’re effective in front of non-technical stakeholders - you can run the meeting, not just attend it.
You move under ambiguity and know which corners are safe to cut in a prototype and which ones aren’t.
You write clearly. Design docs, decision records, and handoff notes are part of the job.
BONUS
Salesforce certifications - Agentforce Specialist, Data Cloud 360 Consultant, Platform Developer II, Application Architect, or Integration Architect.
Production Agentforce deployments with measurable business outcomes.
Data Cloud implementation experience beyond basic grounding.
Open-source contributions to MCP servers, agent tooling, or the Salesforce developer ecosystem.
Prior consulting, solutions engineering, or forward-deployed work at a product company.
Experience with evaluation frameworks for LLM applications.