Job Description

Job Description

Roles & Responsibilities :

This role defines the enterprise framework for secure, ethical, explainable, and compliant AI. Based in the AI CoE, the consultant sets Responsible AI policies, explainability and governance standards, AI risk classification, and regulatory readiness across the AI lifecycle - advising engineering, product, legal, and business teams, and orchestrating security testing. The mandate is to make AI trustworthy and auditable.

Key responsibilities

Responsible AI Framework
- Define and maintain Responsible AI principles, policies, and implementation guidelines - covering fairness, transparency, accountability, privacy, safety, robustness, and human oversight.
- Create Responsible AI assessment templates for AI and GenAI use cases.
- Define AI risk-classification approaches based on impact, data sensitivity, autonomy, user exposure, and regulatory relevance.
- Support review boards and governance forums for high-risk AI initiatives.

Explainability and Transparency
- Define explainability requirements by business criticality,



risk level, and regulatory expectation - and advise teams on appropriate XAI approaches.
- Establish documentation standards - decision logs, datasheets.
- Support business teams in communicating AI-driven decisions in an accountable, and auditable manner.

AI Security and GenAI Risk - Advisory and orchestration
- Define AI security requirements and standards - model access, prompt security, data-leakage prevention, adversarial robustness, and abuse prevention - as policy and guardrails.
- Establish guidelines for GenAI-specific risks: prompt injection, jailbreaks, insecure tool use, data exfiltration, and unsafe outputs.
- Orchestrate AI security testing and red teaming ensure findings are triaged, tracked, and remediated.
- Advise on AI operational risk, including LLMOps observability and cost/FinOps guardrails.

AI Governance and Regulatory compliance
- Develop AI lifecycle governance cont

📌 AI Governance (India)
🏢 Bosch India
📍 India