A decade of building and defending. Today I run AI security and AI governance at Locus (Ingka / IKEA) for a logistics AI platform powering 1.5B+ deliveries across 30+ jurisdictions. Previously secured 35M+ student records across 33 countries and treated 1 million+ vulnerabilities across enterprise environments. I translate AI risk, prompt injection, and model security into board-level business strategy — and build AI security tools in the open.
CISO and DPO at Locus (Ingka / IKEA), running AI security and AI governance for a logistics AI platform powering 1.5B+ deliveries across 30+ jurisdictions. A decade spanning AI SaaS, healthcare, logistics, edtech, robotics, and cloud-native platforms — translating AI risk and board-level cyber risk into business strategy.
The hardest security problems aren't technical — they're organizational. Brought in to stabilize companies during active breaches, PHI exposure, and ransomware incidents. Built incident response, AI incident response, and 24x7 SOC operations from scratch. Led post-acquisition security due diligence for global enterprises. Every role has started with a hard problem and ended with a resilient organization.
Ships production-grade Python AI security tooling in the open — Verida (AppSec false-positive reduction), Vedha (autonomous AI pentester), Nimantrika (safe outbound automation). Designs AI product architectures from first principles and builds entire AppSec pipelines using open-source tooling. The best AI security leaders understand what they're protecting because they've built it themselves.
"The wise warrior avoids the battle."— Sun Tzu, The Art of War
The best security is invisible. It prevents the incident from ever existing. Proactive architectures, quiet resilience, systems that hold when tested.
AI security, AI governance, and enterprise cyber programs that scale with the business — powered by automation and agentic AI. Click any domain to explore.
Running AI security and AI governance for an enterprise AI platform at scale. LLM security guardrails, prompt injection defense, model supply chain controls, AI DPIA, AI TPRM. Aligned to NIST AI RMF, ISO/IEC 42001, EU AI Act, and OWASP Top 10 for LLMs.
AI-assisted AppSec pipelines at enterprise scale. SAST, SCA, DAST, container, IaC, and secrets scanning — with open-source tooling and agentic AI triage reducing false positives and late-stage vulnerabilities.
CISO and DPO across 30+ jurisdictions. AI DPIAs, cross-border transfers, automated decision-making. HITRUST, SOC 2 Type II, ISO 27001, ISO 27701, HIPAA, GDPR, DPDPA, COPPA.
Red-teaming AI systems and traditional infrastructure. OWASP LLM Top 10, jailbreak and data-extraction testing, indirect prompt injection via RAG. Crisis leadership through PHI breaches, ransomware, and AI incident response.
Zero Trust deployments, AI-augmented SOCs processing millions of EPS, SIEM, EDR/XDR, SASE, UEBA/CASB. Identity-first architecture across global enterprises.
A CISO who ships code. Python AI security tools, MCP integrations, agentic automation for SecOps and GRC. Open-source: Verida, Vedha, Nimantrika.
From bug bounty hunter to AI security CISO — a decade of building and defending across 7 industries and 30+ jurisdictions.
Leading AI security, AI governance, privacy, and compliance for a cloud-native logistics AI platform serving 1.5B+ deliveries across 30+ jurisdictions. Established AI security governance with LLM guardrails, AI DPIAs, and AI TPRM aligned to NIST AI RMF and ISO/IEC 42001. Delivered SOC 2 Type II. Led post-acquisition security due diligence for Ingka Group (IKEA), driving remediation of 70,000+ SAST/SCA/OSS vulnerabilities (current-role scope). Lean team of 5, $1M budget. Reduced late-stage vulnerabilities by 96%.
Led IT, Security, Privacy, and Compliance for a global edtech SaaS platform supporting 35M+ student records across 33 countries. Built the entire AppSec pipeline using open-source tooling (Semgrep, Trivy, OWASP ZAP, SonarQube) — eliminating enterprise licensing. Secured AI-enabled IoT ecosystem across 600+ schools with 4,500+ smart classrooms powered by 15,000+ customized devices.
Established global security and privacy across 100+ countries, 2,000+ employees, and 5,000+ contracted teachers. Led security integration during Byju's acquisition, aligning strategy with 300% annual growth. Deployed company-wide Zero Trust architecture. First-time ISO 27001 & ISO 27701 certifications enabling 50+ enterprise B2B deals with global banks, Big4, and Big Tech.
Appointed by Group Chairman to stabilize the organization during a major security crisis — managing a large-scale PHI breach and ongoing ransomware incidents. Built and led global security teams across 3 geographies (25+ professionals, 24x7 SOC). Secured OT/IoT environments for pharmaceutical robotics (30+ robotic platforms, 1,000+ sensors per system). Achieved HITRUST i1, SOC 2 Type II, ISO 27001, ISO 27701, and HIPAA.
Enterprise security governance and architecture across TCS data centers, cloud environments, and critical infrastructure — including the EKA supercomputer used for Indian defense and space research. Key role in corporate security modernization across 1,000+ remote offices. Enterprise incident response for critical infrastructure environments.
Looking for an AI security leader who's been in the trenches and speaks the boardroom's language?
Let's Talk LinkedInOpen-source AI security tools, agentic AI pentesting, and frameworks — built in the open.
Open-source AppSec correlation engine reducing SAST/SCA false positives through reachability analysis and AI-assisted triage. Designed for CI/CD.
Fork of Shannon — an autonomous AI pentester that plans, executes, and reports. Covers OWASP Top 10 for LLMs and traditional web/API pentesting.
Human-mimicking LinkedIn outbound automation. Safe-mode defaults, full jitter, no 24/7 ops — built to protect the account as the load-bearing asset.
Proprietary methodology for building AI products at scale. 9-phase lifecycle, 7 quality gates, 8 development agents. Nyaya AI is the reference implementation.
Security tools should be transparent, community-driven, and accessible.
Short answers to the questions boards, recruiters, and engineers ask most often.
An AI security CISO owns the governance, risk, and controls for AI systems — from model supply chain and prompt injection defense to AI DPIA, AI TPRM, and NIST AI RMF / ISO 42001 alignment. Dr. Devam Shah leads this function at Locus (IKEA), running AI security governance, LLM security reviews, and AI red teaming alongside traditional CISO responsibilities.
AI governance combines policy (acceptable use, data classification, model inventory), controls (input/output guardrails, prompt injection testing, PII redaction, rate limiting), and oversight (AI DPIA, AI TPRM, human-in-the-loop for high-risk decisions). At Locus, Dr. Shah established AI governance aligned to NIST AI RMF and ISO/IEC 42001, with automated guardrails across AI-assisted development pipelines.
MLSecOps extends DevSecOps to the ML/LLM lifecycle — covering training data integrity, model provenance, adversarial robustness, prompt injection, model theft, and inference-time abuse. It adds AI-specific controls on top of SAST, SCA, DAST, and secrets scanning. Dr. Shah builds MLSecOps pipelines using open-source tooling and custom AI guardrails.
AI red teaming tests models and AI-powered applications against prompt injection, jailbreaks, data extraction, model evasion, and indirect injection via RAG sources. Vedha — Dr. Shah's open-source autonomous AI pentester — automates this across the OWASP Top 10 for LLMs and extends into traditional pentesting workflows.
A DPO handling AI systems must run AI DPIAs, assess legal basis for training and inference, manage cross-border transfers under GDPR and DPDPA, track automated decision-making under EU AI Act high-risk categories, and coordinate with the CISO on model security. Dr. Shah operates as both CISO and DPO — making AI privacy and AI security a unified function rather than two siloed ones.
Dr. Shah advises boards and executive teams on AI security strategy, AI governance program design, AI risk translation, and AI compliance readiness (NIST AI RMF, ISO/IEC 42001, EU AI Act). Contact via email or LinkedIn to discuss.
Lessons from the trenches — AI risk, LLM defense, board-level security strategy, and building security tools in the open.
Most enterprises treat AI governance as a checkbox exercise. After building AI governance programs across three companies, here's what actually works — and what's theater.
Read moreStop treating prompt injection like XSS. It's a fundamental trust boundary violation that requires architectural solutions, not input sanitization.
Read moreEnterprise AppSec doesn't require enterprise licensing. Here's the open-source pipeline that outperformed six-figure commercial tools at Locus.
Read moreYour AI security & AI governance advisor. Ask about LLM defense, AI DPIA, AI red teaming, or how I'd handle a specific AI risk.
Whether it's a board that needs AI risk translated into business language, an AI platform that needs securing, or a crisis that needs steady hands — I've been there. AI security strategy · AI governance · board advisory.