Vaibhav Bhardwaj — DevOps Engineer

I operate at the critical intersection of enterprise identity infrastructure and AI safety. As a CAISP & CISM-certified AI Security Engineer with 18 years of experience, I bring a rare combination of deep Privileged Access Management (PAM), Identity & Access Management (IAM), and cutting-edge AI security expertise. Currently at GIC Singapore, I design, evaluate, and stress-test the enterprise AI stack — from LLM Gateways and AI Guardrails to secure agentic workflows and Model Context Protocol (MCP) implementations. My focus is building production-grade defenses for the next generation of autonomous AI systems. I specialize in adversarial red-teaming, zero-trust agent architectures, and identity-aware policy enforcement that protects organizations against prompt injection, tool abuse, data exfiltration, and other emerging AI threats. Key Highlights AI Security Infrastructure: Led a comprehensive 7-vendor RFI and architectural evaluation for LLM Guardrails and Gateways, aligned to OWASP Top 10 for LLMs and MITRE ATLAS. Shortlisted top 3 vendors for PoC within 8 weeks. Adversarial Red Teaming: Designed and executed scalable PoC test harnesses to stress-test guardrails against prompt injection, jailbreaks, indirect injections, SSRF/tool abuse, and data exfiltration. Quantified block rates, FP/FN ratios, and latency overhead (220–380ms). Zero-Trust Agentic Workflows: Architected identity-aware pipelines using AWS AgentCore, PingAuthorize (ABAC), and dynamic Step-Up MFA. Enforced strict "Chinese Wall" isolation between Public and Private Market domains while maintaining least-privilege access at every step. Endpoint & Developer Tool Security: Conducted offensive red-team assessments on GitHub Copilot and Claude Desktop, identifying real-world exfiltration paths and implementing hardened policies, network controls, and SIEM detections. MCP Security: Engineered secure middleware for Model Context Protocol environments, delivering multi-tenant isolation, cryptographic verification, content filtering, and secrets redaction. My foundational expertise in PAM, Ping Identity suite and cloud-native secrets management allows me to approach AI security challenges with a deep understanding of identity — widely considered the new perimeter for agentic AI systems. I care deeply about the responsible adoption of AI in regulated industries and always enjoy exchanging ideas with fellow security engineers, architects, and AI governance leaders. Open to conversations on AI security, agentic workflows, LLM defense strategies, or potential collaborations.

Stackforce AI infers this person is a Cybersecurity expert specializing in AI security and identity management for enterprise solutions.

Location: Singapore, Singapore

Experience: 16 yrs 9 mos