Agentforce and Salesforce AI Specialist

Agentforce and Salesforce AI Specialist

Location: Chennai

Employment type: Full Time

Experience: 4+ years in Salesforce delivery or applied AI engineering, including practical implementation and evaluation of Salesforce automation and AI agent use cases.

About Teceze

Teceze is an IT services company supporting clients across Digital Workplace Services, Cloud and Infrastructure, Cyber Security, Staff Augmentation and Global Procurement Services. This role contributes to client Salesforce consulting and delivery, working with business stakeholders and technical teams to implement, improve and support enterprise applications.

Role purpose

Teceze is seeking an Agentforce and Salesforce AI Specialist to build useful, controlled AI agents for client sales, service and internal operations. You will translate business workflows into agent behaviour, connect agents to reliable data and approved actions, and prove that solutions are safe, effective and supportable before production use.

Key responsibilities

• Identify suitable agent use cases with business owners, documenting intended users, success criteria, failure modes and boundaries of autonomous action.
• Configure agent instructions, topics or equivalent routing constructs, actions and conversation flows to support clearly defined business tasks.
• Build reusable prompt templates and connect agent actions to Flow, Apex and approved APIs with explicit input and output validation.
• Select and prepare grounding sources, retrieval approaches and data access so responses reflect relevant, authorised and sufficiently current information.
• Design escalation, human handoff and approval steps for uncertain responses, sensitive requests and actions with significant business consequences.
• Develop evaluation datasets and adversarial test cases covering factual accuracy, task completion, misuse, prompt injection and unauthorised data exposure.
• Test multi-step behaviour, exception handling and tool failures; compare alternative designs against quality, response-time and consumption objectives.
• Coordinate deployment, monitoring, version control and rollback with platform and security teams; document operational ownership before release.
• Analyse production feedback and agent outcomes, prioritise corrections and maintain regression tests as prompts, actions and source data evolve.
• Demonstrate solutions to clients, train administrators and service teams, and provide realistic estimates of implementation effort and ongoing support.

Required skills and experience

• Practical experience building and testing Agentforce solutions, supported by a working demonstration or documented delivery evidence.
• Strong Salesforce platform fundamentals, including Flow, permissions, data models and the implications of executing actions in a user or system context.
• Understanding of large language model behaviour, grounding, retrieval-augmented generation and the limits of prompt-based controls.
• Ability to define measurable evaluations rather than relying only on conversational demonstrations or subjective output quality.
• Working knowledge of APIs and integration patterns, with ability to collaborate on Apex actions and investigate failures.
• Clear communication about AI capabilities, limitations, operational risks and where human review remains necessary.

Preferred qualifications

• A current Salesforce Agentforce credential, alongside relevant platform administration or development credentials.
• Experience with Data Cloud or Data 360, search and retrieval systems, Python, or automated AI evaluation tooling.
• Experience deploying agents into customer service workflows with measurable quality and escalation

Key deliverables

Agent use-case specifications, prompts and action configurations, grounding design, evaluation datasets, test evidence, monitoring dashboards and handoff runbooks.

Measures of success

• Agents meet agreed task-completion and response-quality thresholds on representative evaluations before release.
• Access boundaries, approval steps and escalation paths perform as designed, with actionable monitoring of failures and unexpected behaviour.
• Business outcomes improve against a documented baseline while operating costs and response times remain within agreed limits.