Lead - AI Unit Senior Lead - AI Unit

Location Name: Pune Corporate Office - Mantri

Job Purpose

This role leads the domain strategy, reasoning design, and orchestration blueprint for Agentic AI use cases across the Risk, Underwriting, HR, and RCU (Risk Control Unit) verticals. The position is responsible for translating credit policies, underwriting rules, HR process flows, risk frameworks, governance structures, investigations logic, and compliance checkpoints into machine-interpretable domain structures for autonomous agents. The role is central to driving trustworthy, auditable, domain-grounded reasoning and will anchor collaboration between business, risk, HR, compliance, and the Agentic AI tech teams to ensure accurate, safe, and production-ready agentic systems.

Duties And Responsibilities

Lead the end-to-end domain modelling, reasoning architecture, and orchestration logic for Agentic AI use cases across Risk, Underwriting, HR, and RCU.

• Translate complex policy documents, HR guidelines, risk frameworks, underwriting rules, compliance playbooks, and investigation workflows into structured reasoning frameworks, rule engines, and decision flows.
• Define agent-level guardrails, constraint boundaries, fallback behaviors, escalation routes, and risk thresholds for autonomous actions.
• Architect and maintain domain knowledge bases, including credit heuristics, personnel policies, case-handling pathways, risk scoring logic, and RCU checks.
• Design thought-logging schemas, reasoning trace structures, and interpretability checkpoints to support audit, compliance, fairness, and governance reviews.
• Collaborate closely with the Agentic AI tech team to integrate domain logic into agent frameworks, multi-agent orchestration models, and system APIs.
• Drive multi-layer testing, scenario evaluation, bias checks, and reasoning audits to ensure agent reliability, fairness, and alignment to corporate standards.
• Partner with Credit, Risk, HR, Payroll, Talent, Compliance, Fraud/RCU, and Investigations teams to continuously refine domain accuracy.
• Identify and prioritize new high-value use cases across the four verticals, shaping the agent roadmap.
• Oversee documentation quality, ensuring domain modules, reasoning libraries, orchestration guides, and logic maps meet governance standards.
• Serve as the domain authority for internal stakeholders, ensuring all agent behaviors remain compliant with policy, regulatory norms, and ethical AI guidelines
  
  
  

• Determine logging granularity, interpretability depth, and checkpoints required for governance.
• Approve fallback and escalation flows for sensitive cases in underwriting, HR, risk, and investigations.
• Prioritize domain enhancements, new initiatives, and cross-functional integrations.
• Frame evaluation frameworks for agent reasoning: fairness, accuracy, stability, compliance alignment
  
  
  

• Translating implicit expert reasoning into explicit, machine-executable decision frameworks.
• Managing uncertainty, ambiguity, and contradictory signals in domain data without oversimplifying. Interpretability, Logging, and Observability
• Designing logging schemas that support post-hoc reasoning audits without creating overhead or data bloat.
• Balancing performance with observability, especially in production settings. Strategic Autonomy vs Control
• Deciding when agents should self-direct vs. escalate decisions to humans.
• Embedding guardrails that are both domain-relevant and adaptive under evolving agent capabilities. Feedback Loop Complexity
• Designing multi-layered feedback systems: self-reflection ? user feedback ? domain overrides ? model fine-tuning.
• Managing feedback prioritization: what gets logged, analyzed, and applied—and what gets ignored
  
  
  

• 8–12 years of experience in Risk, Underwriting, HR, Compliance, RCU, or related governance domains, with exposure to AI or automation.
• Strong understanding of credit policy, risk scoring frameworks, HR SOPs, RCU processes, and compliance guidelines.
• Deep familiarity with decision systems, rule engines, case management workflows, and governance processes.
• Working knowledge of Agentic AI frameworks, LLM reasoning, orchestration pipelines, and interpretability best practices (preferred).
• Ability to translate policy-heavy content into structured, auditable, machine-executable logic.
• Excellent stakeholder communication, cross-functional alignment skills, and domain documentation ability.
• Strong analytical, problem-solving, and risk-awareness capabilities.