Agentic AI Security Specialist

DELTACLASS TECHNOLOGY SOLUTIONS LIMITED; Greater London; Sur place; GBP 120 000 - 180 000; Plein temps

Résumé du poste

DELTACLASS TECHNOLOGY SOLUTIONS LIMITED is seeking an Agentic Threat Modeling & Secure Engineering specialist to design secure architectures for multi‑agent systems, enforcing least privilege and sandboxing.

You will build real-time guardrails, implement Advanced Threat Protection boundaries, and develop detection logic to stop prompt injections and data exfiltration, while collaborating with DevOps and Agile teams.

Qualifications

• Experience in threat modeling for multi-agent systems and secure architecture design.
• Knowledge of least privilege principles and sandboxing for autonomous agents.
• Ability to implement real-time guardrails and threat protection boundaries.

Responsabilités

• Perform specialized Threat Modeling on multi-agent orchestrations and vulnerabilities.
• Architect secure execution environments and sandboxes with strict privilege controls.
• Implement real-time guardrails to intercept malicious prompts and data exfiltration
• Execute Offensive Simulation & Detection Engineering to identify anomalous agent behaviors.
• Develop Response Engineering to isolate or throttle rogue agent threads without disruption.

Connaissances

Threat modelling

Security engineering

Python

Go

DevOps

Agile

Outils

Jira

Confluence

LangChain

AutoGen

CrewAI

Description du poste

1. Agentic Threat Modeling & Secure Engineering

• Perform specialized Threat Modeling on multi-agent orchestrations, focusing on tool-use vulnerabilities, autonomous execution manipulation, goal-jacking, and cascading agent failures.
• Architect secure execution environments (Security Engineering) and sandboxes to isolate autonomous agents, ensuring strict enforcement of the principle of least privilege.
• Implement real-time proxy guardrails and Advanced Threat Protection boundaries to intercept malicious prompts, indirect prompt injections, and data exfiltration vectors.

2. Offensive Simulation & Detection Engineering

• Execute Offensive Security Testing specifically mimicking adversarial machine learning tactics against agent reasoning loops, vector databases, and memory layers.
• Design custom Detection Engineering logic to identify anomalous agent behaviors, API misuses, and malicious intent vectors before downstream execution occurs.
• Develop automated orchestration pathways (Response Engineering) to instantly isolate, throttle, or terminate rogue agent threads without disrupting core systems.

3. Core Language & Automation Engineering

• Write production-grade security proxies, wrapper interfaces, and low-latency validation controls using Python and the Go Programming Language.

Secondary Roles & Responsibilities

• Conduct proactive Cyber Threat Hunting exercises to uncover hidden signs of systemic poisoning, persistent agent manipulation, or silent model exploitation.
• Perform advanced Digital Forensics on complex agent memory logs, tracking step-by-step reasoning patterns and API execution loops following anomalous events.

2. Governance, Delivery, & Enablement

• Embed security guardrails smoothly into production deployment loops using DevOps pipelines (CI/CD) and automated testing frameworks.
• Operate within high-velocity teams using Agile Methodology & Tools (e.g., Jira, Confluence) to cleanly deliver tasks inside sprint timelines.
• Maintain a mindset of Customer Centricity, ensuring that heavy security layers and model guardrails do not negatively disrupt the user experience.
• Commitment to Continuous Learning, tracking the fast-evolving landscape of agentic frameworks (e.g., LangChain, AutoGen, CrewAI) and modern security benchmarks (MITRE ATLAS, OWASP Top 10 for LLMs).