AI Trust and Governance Architect

Infosys Limited seeks an AI Trust and Governance Architect in Bangalore for its Infosys Quality Engineering service line. The role designs AI assurance architecture and platforms to evaluate AI systems, manage risk, and support responsible, reliable deployment.

Responsibilities

• Architect assurance, evaluation, and benchmarking frameworks for LLMs, agents, and retrieval-augmented generation (RAG), covering functional, non-functional, and risk dimensions. Define patterns for bias detection, explainability, safety validation, and reusable business, risk, and reliability assurance components.
• Design security, privacy, prompt-injection, adversarial, red-teaming, threat-simulation, and chaos-style validation. Establish model-usage governance, auditability, traceability, and compliance mechanisms, while addressing resilience, fault tolerance, and observability.
• Build platforms for automated AI testing, reporting, and insights; integrate quality gates with CI/CD pipelines and work with quality-engineering teams to embed assurance in the software development lifecycle. Mentor teams on AI risk identification and mitigation.
• Partner with product and engineering teams on assurance opportunities and roadmaps. Support client workshops, RFPs, and solution presentations, and explain AI concepts to business audiences.

Required qualifications

• Bachelor of Engineering. The posting lists 8–15 years of work experience; its must-have requirements specify 13+ years in software engineering, including 3+ years in AI, with strong architecture ownership.
• Hands-on expertise in AI/ML systems, LLM evaluation, and assurance frameworks; experience with AI red teaming, model risk management, or AI audit tooling; knowledge of Responsible AI, AI risks, and governance; experience in security and adversarial testing and reliability engineering; proficiency in Python, automation frameworks, and cloud platforms.

Tools and additional skills:

• Listed tooling includes PromptFoo, DeepEval, custom LLM evaluation harnesses, prompt and retrieval validators, agent workflow evaluators, Fairlearn, SHAP, toxicity and bias scanners, PyRIT, Garak, Galileo, Langfuse, Arize, Evidently, and telemetry dashboards.
• Good-to-have qualifications include AI regulatory or compliance knowledge, performance or chaos engineering and resilience testing, and contributions to platforms, frameworks, or standards. Other listed preferred areas include COBIT assurance, governance, risk and compliance, audits, workflows, data governance, conversational AI platforms, RAG, AI/ML solution architecture, Responsible AI, and prompt engineering.