Agentic AI SME

About The Job:

• Agentic AI Subject Matter Expert (SME)
• The Agentic AI SME is responsible for converting approved architecture and batch-analysis objectives into a controlled, explainable, and testable Agentic AI capability.
• This role defines agent workflows, grounding strategies, tool integrations, evaluation frameworks, and human-review controls required to ingest batch-processing evidence, identify dependencies, and generate optimization recommendations.
• The position acts as the technical bridge between Agentic AI platforms, data pipelines, architecture teams, and Mainframe Batch SMEs.
• The focus is on reliable analysis, evidence traceability, transparent uncertainty management, and safe human validation rather than autonomous production changes.
• Primary accountability includes quality, explainability, reproducibility, governance, evidence lineage, confidence scoring, evaluation results, and human-in-the-loop controls.

Essential Job Functions:

• Translate approved use cases and batch-analysis objectives into agent roles, workflows, tool interfaces, prompts, policies, knowledge sources, and decision boundaries.
• Design analysis workflows for ingesting and reconciling scheduler metadata, JCL, logs, configurations, calendars, resource data, and other approved batch artifacts.
• Partner with data and platform engineers to define schemas, normalization standards, graph structures, lineage requirements, versioning, exception handling, and repeatable ingestion processes.
• Develop or configure agent capabilities for dependency discovery, critical-path analysis, bottleneck detection, resource utilization analysis, redundant-job identification, and optimization recommendation generation.
• Ensure outputs are grounded in approved source evidence and that all findings are traceable to source inputs, transformations, prompt versions, and analysis context.
• Implement confidence scoring, exception handling queues, and escalation mechanisms for incomplete, contradictory, or low-quality evidence.
• Define evaluation datasets, testing scenarios, acceptance criteria, and validation frameworks in collaboration with Mainframe Batch SMEs.
• Evaluate solution accuracy, completeness, consistency, relevance, explainability, and false-positive rates.
• Implement human-in-the-loop controls ensuring AI outputs are validated before being accepted as findings.
• Support platform architecture decisions with AI functional and non-functional requirements, security requirements, model governance, orchestration, and knowledge-management considerations.
• Support platform readiness activities including model configuration, orchestration setup, logging, monitoring, access controls, testing, and validation.
• Create reusable assets such as agent specifications, prompt libraries, governance policies, evaluation packs, model cards, runbooks, and traceability documentation.
• Clearly communicate AI capabilities, limitations, assumptions, risks, and recommendations to stakeholders, architects, project managers, and SMEs.

Role Boundaries & Interfaces

• Owns:
  • Agentic workflow and analysis design
  • Grounding, traceability, and AI evaluation
  • Prompt, tool, and knowledge governance
  • AI quality and readiness evidence
• Partners On:
  • Batch-domain rules and validation
  • Platform, data, and security engineering
  • Architecture approval and operational readiness
  • Client workshops and acceptance activities
• Does Not Own:
  • Final mainframe-domain sign-off
  • Autonomous production changes
  • Approval of business or regulatory decisions
  • Treating unvalidated AI outputs as facts

Key Deliverables

• Agentic Analysis Design: Agent roles, workflows, tools, grounding frameworks, controls, and decision boundaries.
• Data & Evidence Specifications: Schemas, lineage, normalization standards, validation requirements, and exception management frameworks.
• Prompt / Policy / Tool Library: Version-controlled prompts, policies, tools, and knowledge-source configurations.
• Evaluation & Assurance Pack: Golden test cases, negative tests, quality measures, evaluation results, and remediation actions.
• Analysis Outputs: Dependency analysis, bottleneck identification, resource utilization insights, and optimization recommendations with supporting evidence.
• Operational Documentation: Runbooks, monitoring procedures, logging strategies, support processes, and recovery guidance.
• Knowledge Transfer: Training, walkthroughs, and reusable guidance for project teams and business stakeholders.

Qualifications:
Required Skills & Experience

• Hands-on experience designing and implementing enterprise Agentic AI, Generative AI, LLM, or advanced analytics solutions.
• Experience with multi-agent orchestration, tool calling, workflow automation, and AI-assisted decision-support systems.
• Strong expertise in Retrieval-Augmented Generation (RAG), embeddings, structured data extraction, dependency analysis, knowledge graphs, and evidence-grounded AI.
• Deep understanding of prompt engineering, model governance, knowledge governance, version control, guardrails, and auditability.
• Experience defining evaluation frameworks and metrics such as accuracy, consistency, hallucination rates, unsupported claims, and task completion quality.
• Hands-on experience integrating AI solutions with structured and unstructured enterprise data sources.
• Strong understanding of secure cloud architectures, IAM, encryption, CI/CD, secrets management, monitoring, and environment separation.
• Ability to explain AI behavior, risks, limitations, and outcomes to both technical and non-technical stakeholders.
• Strong collaboration skills across architecture, security, data engineering, platform engineering, and business teams.

Preferred Skills

• Experience with Kyndryl Agentic AI Framework or similar enterprise Agentic AI platforms.
• Experience with AWS-based AI and data services.
• Knowledge of mainframe environments, batch scheduling systems, JCL, and operational analytics.
• Experience with graph databases, knowledge graphs, anomaly detection, time-series analysis, and workload optimization.
• Understanding of banking AI governance, model risk management, data residency, compliance, and regulated environments.

Success Profile

• AI-generated outputs are reproducible, evidence-based, explainable, and understandable to domain SMEs and client stakeholders.
• Quality is measured against predefined test cases and evaluation criteria rather than demonstrations alone.
• Low-confidence findings are identified, flagged, and routed for human review.
• The Agentic AI solution produces decision-ready insights without performing unauthorized production actions.
• All recommendations are supported by traceable evidence, confidence scores, assumptions, risks, and validation requirements.

Scope Guardrail

• The solution is designed to analyze, evaluate, and recommend.
• Implementation of batch optimization changes and autonomous production actions remain out of scope unless separately approved through formal change control processes.

How to Apply:Interested candidates are encouraged to respond/submit their updated resumes, and for additional job opportunities, please visitJobs In India – VARITE.

Unlock Rewards:Refer Candidates and Earn.

If you're not available or interested in this opportunity, please pass this along to anyone in your network who might be a good fit and interested in our open positions. VARITE offers a Candidate Referral program, where you'll receive a one-time referral bonus based on the following scale if the preferred candidate completes a three-month assignment with VARITE.

Experience Level Bonus Referral:

About VARITE: VARITE is a global staffing and IT consulting company providing technical consulting and team augmentation services to Fortune 500 Companies in USA, UK, CANADA and INDIA. VARITE is currently a primary and direct vendor to the leading corporations in the verticals of Networking, Cloud Infrastructure, Hardware and Software, Digital Marketing and Media Solutions, Clinical Diagnostics, Utilities, Gaming and Entertainment, and Financial Services.

Equal Opportunity Employer:
VARITE is an equal opportunity employer. We celebrate diversity and are committed to creating an inclusive environment for all employees. We do not discriminate based on race, color, religion, sex, sexual orientation, gender identity or expression, national origin, age, marital status, veteran status, or disability status.