AI Product Experience Compliance

Role- AI Product Experience Compliance

Location- Noida & Hyderabad

Experience- 14-21 Years

Role Summary:

Define the AI harnessing and compliance specification — safety, guardrails, explainability, auditability — and build the validation mechanism the CTO org's builds are tested against. Secondary but connected scope: identify where user-experience friction (e.g., excessive steps to complete a task, unclear agentic responses) creates compliance or safety risk — specifically where friction causes users to bypass intended safeguards, not UX quality in general.

Key Responsibilities:

Core (compliance/safety):

\n
• Specify AI guardrails: prompt/response policy, PII handling, refusal and escalation behavior
\n
• Build automated compliance checks and red-team test suites into the validation harness
\n
• Maintain the audit trail and evidence pack for internal and customer audits
\n
• Track applicable AI regulation (EU AI Act, sector-specific rules) and translate into engineering requirements, in partnership with legal/compliance review — this role proposes interpretations, it does not have unilateral legal authority
\n
• Validate delivered AI behavior against the harnessing spec at each release
\n
• Evaluate model outputs for bias/fairness issues relevant to the applicable regulatory framework
\n

Secondary (experience-risk signal):

\n
• Capture specific UX-compliance-risk signals — e.g., number of steps/clicks to complete a task, points where users abandon or override a safety confirmation, ambiguity in agentic responses that leads to user error — and flag where these create compliance exposure, distinct from general UX quality feedback
\n

Must-Have Requirements:

\n
• 12+ years full-stack engineering with hands-on LLM application work
\n
• Has implemented guardrails, content filtering, or model-output validation in a production system — be ready to describe the architecture and a specific failure it caught
\n
• Deep understanding of prompt injection, jailbreak techniques, and data-leakage failure modes
\n
• ML evaluation experience, including bias/fairness evaluation methodology
\n
• Regulated-industry exposure (finance, healthcare, or similar) with direct experience operating under an external audit
\n
• Hands-on experience with named red-teaming/guardrail/evaluation tooling — specify which from: Garak, PyRIT, Promptfoo, NeMo Guardrails, Llama Guard, Guardrails AI, TruLens, Ragas, or equivalent
\n
• Can read regulatory text (EU AI Act or sector-specific) and produce draft test cases, with the explicit understanding that a compliance/legal function reviews the final interpretation
\n

Preferred:

\n
• UX research or product analytics background (task-completion metrics, friction analysis) — genuinely useful for the secondary responsibility, but should not be must-have alongside the security/compliance bar
\n
• Experience with formal AI governance frameworks (NIST AI RMF, ISO/IEC 42001)
\n
• Prior work directly supporting an external regulatory audit (not just internal compliance)
\n