yoinka

AI Security Engineer

Allstate

RemoteUS - RemoteFull TimeMid
Sign in to applyVerified
Location
US - Remote
Employment
Full Time
Work model
Remote
Level
Mid
Posted

Skills

CI/CDJavaLLMPythonREST

About this role

At Allstate, great things happen when our people work together to protect families and their belongings from life’s uncertainties. And for more than 90 years, our innovative drive has kept us a step ahead of our customers’ evolving needs. From advocating for seat belts, air bags and graduated driving laws, to being an industry leader in pricing sophistication, telematics, and, more recently, device and identity protection.

Job Description

Allstate is seeking an AI Security Engineer to help build and operate an always-on adversarial testing capability for AI-enabled and agentic systems. In this role, you will integrate commercial, open-source, and Allstate-internal adversarial testing technologies into Allstate environments, connect them to AI-enabled applications, and build the automation, telemetry, and data capabilities needed to execute testing reliably at enterprise scale. You will work closely with engineers designing new adversarial techniques to turn successful strategies into repeatable capabilities that can continuously identify and measure AI system weaknesses. This is a hands-on software and platform engineering role suited for engineers who enjoy integrating emerging technologies into complex enterprise environments.

Key Responsibilities

Integrate commercial, open-source, and Allstate-built AI agentic evaluation and red-team technologies into Allstate environments using approved enterprise engineering and security patterns. Connect testing systems to AI products through APIs, agent and tool protocols, voice interfaces, and other application-specific integration patterns. Turn attack strategies into reliable, configuration-driven test capabilities that can be reused across applications and environments. Develop APIs, microservices, orchestration components, and automated triggers supporting continuous and large-scale adversarial testing. Build secure authentication, authorization, and credential-management patterns that allow automated testing to operate without manual intervention. Capture and persist test evidence including requests, responses, conversation history, tool interactions, configuration, grading results, and other relevant telemetry. Integrate test results with reporting, observability, compliance, and engineering workflows. Build CI/CD, monitoring, logging, tracing, alerting, and operational practices for the adversarial testing platform. Essential Skills Professional software engineering experience building and operating enterprise applications, services, or platforms. Strong proficiency in Python and/or Java. Experience designing and integrating REST APIs, microservices, asynchronous workflows, or event-driven systems. Experience integrating third-party, vendor, or open-source technologies into enterprise environments. Experience deploying applications using cloud, container, or enterprise PaaS technologies. Experience with CI/CD, automated testing, source control, configuration management, and modern software delivery practices. Understanding of authentication and authorization, service identities, secrets management, network controls, and secure application integration. Experience building observable services using logs, metrics, tracing, alerts, and operational dashboards. Strong problem-solving and troubleshooting skills across application, integration, deployment, and infrastructure boundaries.

Preferred Skills

Experience with AI/LLM evaluation, red-team, security testing, or quality-assurance frameworks. Experience with tools such as Coval, Promptfoo, PyRIT, Garak, Inspect AI, or similar technologies. Experience with agentic AI systems, including tool-calling agents, MCP, multi-agent systems, conversational agents, or voice agents. Familiarity with LLM evaluation, LLM-as-a-judge techniques, AI guardrails, or AI observability. Familiarity with AI security risks such as prompt injection, sensitive-data disclosure, excessive agency, memory poisoning, or insecure tool use. The hiring

Listing verified . Applications go through the company's official careers site.

← Back to Yoinka