Senior AI Security Engineer

About the job

Company We partner with enterprises to advise, build, secure, and operationalize AI systems at scale. Our focus is on developing Generative AI (Gen AI), Agentic AI, and Reinforcement Learning-driven systems, while embedding security, governance, and risk controls directly into AI workflows. We enable organizations to safely deploy LLMs, autonomous agents, and adaptive decisioning systems in regulated, mission-critical environments. Job Description As a Senior AI Security Engineer (Gen AI, Agentic AI & Reinforcement Learning), you will lead the design and implementation of secure, scalable, and adaptive AI systems, including LLM-based applications, agentic workflows, and RL-driven decision engines. This role goes beyond traditional security—you will build intelligent, self-improving security review systems using agentic frameworks (Lang Graph, Lang Chain, Lang Smith) and reinforcement learning techniques to continuously enhance AI risk evaluation, policy enforcement, and approval workflows. Key Responsibilities Gen AI, Agentic AI & RL Security Architecture Design and secure LLM, RAG, multi-agent, and RL-driven systems Implement security controls for: • Autonomous decision-making agents • RL-based adaptive systems • Tool-using and API-integrated agents • Ensure safe exploration and bounded behavior in RL environments Agentic AI + Reinforcement Learning for Security Automation (Core Focus) Build agentic AI pipelines using: • Lang Graph → multi-step, stateful security workflows • Lang Chain → LLM orchestration and tool integration • Lang Smith → observability, tracing, and evaluation Develop RL-enhanced security agents that: • Learn from past approval decisions • Optimize risk scoring and classification over time • Continuously improve policy enforcement accuracy Implement feedback loops (human-in-the-loop + automated) to train: • Risk evaluation agents • Compliance validation agents Automate end-to-end intake → evaluation → approval pipelines for Gen AI and Agentic AI use cases Reinforcement Learning Implementation & Governance Design and implement RL models for adaptive security decisioning • Policy optimization • Risk-based prioritization • Dynamic access control adjustments Apply safe RL techniques: • Reward shaping aligned with compliance and security policies • Constraint-based RL (safe exploration boundaries) Monitor and mitigate risks such as: • Reward hacking • Unsafe policy learning • Drift in learned behaviors Integrate RL models into AI governance workflows for continuous improvement AI Risk, Governance & Compliance Translate frameworks such as: • NIST AI RMF • EU AI Act • OWASP Top 10 for LLMs into automated, adaptive controls Build dynamic risk scoring systems enhanced by RL: • Adversarial Risk Score • Model Drift Index • Policy Compliance Confidence Score Generate real-time AI risk heat maps and approval recommendations Implement policy-as-code + policy-learning systems Security Assessment & Red Teaming Conduct AI/LLM/RL system security assessments Perform red teaming across: • Prompt injection scenarios • Agent tool misuse • RL policy exploitation Evaluate vulnerabilities in: • RAG pipelines • Multi-agent coordination • RL training environments AI/ML Lifecycle & LLMOps/RLOps Security Secure the full lifecycle: • Data ingestion, labeling, and validation • Model training (LLM + RL) with GPU isolation and sandboxing • Deployment, inference, and continuous learning loops Implement RLOps + LLMOps security controls Ensure: • Model lineage and provenance • Secure feedback loops • Version control for policies and learned behaviors Monitoring, Incident Response & Observability Build AI + RL-aware monitoring systems • Detect anomalies in: • LLM outputs • Agent decisions • RL policy shifts Develop incident response playbooks for autonomous systems Create executive dashboards linking AI + RL risk to business KPIs Data Security & Access Control Implement fine-grained and adaptive access controls Secure: • RAG knowledge bases • Vector databases • RL training datasets Ensure compliance with data privacy and residency requirements Thought Leadership Act as an SME in: • AI Security • Agentic AI systems • Reinforcement Learning security Research emerging risks in: • Autonomous AI systems • Self-improving models • Multi-agent + RL ecosystems Qualifications Required • Bachelor’s degree in Computer Science, Engineering, or related field • 3–5+ years of experience in cybersecurity (application, cloud, or data security) • Strong experience in automation, scripting, and security tool development.