Role description

Job Description

Salesforce Agentforce

Job Summary -

Design, build, and productionize AI agents on Salesforce Agentforce - from the agent's reasoning and tools, through the data that grounds it (Data Cloud, Snowflake), the systems it acts on (via MuleSoft and APIs), to the custom UI users interact with. You translate fuzzy business problems ("help brokers manage deals conversationally") into reliable, testable agents.

Years of experience needed - 7years

Technical Skills:

Proven experience building Salesfroce Agentforce (or comparable LLM agent frameworks): agents/topics/tools, prompt engineering, and grounding.
Strong Salesforce platform depth: Apex, SOQL, the security model, and LWC.
Hands-on Salesforce Data Cloud: ingestion, data modeling (DMOs), identity resolution, and grounding/retrieval for AI.
Working SQL + data modeling skills and the ability to understand a Snowflake (or similar cloud data warehouse) schema and validate data quality.
Real integration experience - MuleSoft or equivalent iPaaS / API platforms - and sound API design instincts.
Solid prompt engineering judgment: grounding, guardrails, evaluation, and an understanding of where deterministic code must back up probabilistic models.

Build the agent
Design Agentforce agents end to end: topics/subagents, planners, routing, session/context variables, and the response contract that governs agent output.
Author and own the agent's tools - invocable actions (Apex, Flows, prompt templates, external services) - including the input/output schemas the LLM reasons over.
Engineer prompts as production artifacts: instructions, grounding, and tool descriptions that steer routing and behavior predictably. Design for LLM non-determinism rather than against it.
Establish evaluation: utterance test sets (should-trigger / should-not), regression checks across topics, and a feedback loop for continuous tuning.

Ground it in data
Model and activate data in Salesforce Data Cloud: ingest streams, map to data model objects (DMOs), build calculated insights, and configure retrievers / search indexes that ground the agent (RAG).
Work fluently with Snowflake: understand warehouse data models, write SQL to explore and validate data, and connect Snowflake to Data Cloud via zero-copy / data sharing so the agent can reason over enterprise data without brittle copies.
Decide what data belongs natively in CRM vs. federated from the lakehouse, and why.

Integrate the systems it touches
Design integrations to source/target systems using MuleSoft (API-led connectivity, system/process/experience APIs) and REST/platform patterns - so the agent can read and write beyond Salesforce safely and within governance.
Handle auth, error handling, idempotency, and rate/governor limits across the boundary.

Surface it in a UI
Build the conversational front end: native Agentforce UI and/or custom UI (Lightning Web Components, and embedded web/React experiences) for richer interactions - record cards, forms, selectors, file upload.
Mobile-first: responsive layouts, touch, intermittent connectivity, real-device testing.

Process Skills:

Excellent written and verbal communication skills

Behavioral Skills:

Resolve issues of projects and explore to alternate designs

Effectively collaborates and communicates with the stakeholders and ensure client satisfaction

Train and coach members of project groups to ensure effective knowledge management activity.

Certification:

Any relevant certification would be added advantage