AI Cybersecurity Engineer

ECS is seeking a AI Cybersecurity Engineer to work in our Arlington, VA office.

We are seeking a skilled AI Cybersecurity Engineer to ensure the secure deployment, monitoring, and optimization of artificial intelligence models across production environments. This role bridges the gap between AI model development and operational systems, integrating models into enterprise applications, APIs, and cloud or on-premises infrastructure. The engineer will build observability frameworks for real-time and historical model health, detect and mitigate data drift, and apply secure-by-design principles to safeguard AI assets. This position is ideal for candidates experienced in AI integration, cybersecurity, and system observability who can operate at the intersection of data science, DevSecOps, and compliance engineering.

Responsibilities:

• Integrate AI/ML models into enterprise applications (e.g., web, mobile, IoT) using APIs such as REST or gRPC and serving frameworks like TensorFlow Serving or AWS SageMaker.

• Design and implement real-time and historical dashboards using Grafana, Kibana, or Plotly to monitor model health indicators such as latency, accuracy, and utilization.

• Implement automated pipelines using tools such as Evidently AI or Weights & Biases to detect data drift and model degradation, generating alerts for rapid remediation.

• Logging and Tracing: Configure comprehensive logging and tracing systems using ELK Stack, OpenTelemetry, or LangSmith to capture AI events, system traces, and error logs for debugging, auditing, and compliance.

• Apply secure-by-design and adversarial resilience practices to safeguard AI models from threats such as data leakage, prompt injection, or model inversion attacks. Utilize frameworks such as the Adversarial Robustness Toolbox (ART).

• Optimize model inference performance through techniques like quantization or edge deployment while ensuring compatibility with hybrid and cloud infrastructures (AWS, Azure, or on-premises).

• Partner with data scientists, MLOps, and DevSecOps teams to align model integration with infrastructure, security, and business requirements.

• Conduct end-to-end testing and validation of integrated AI systems, including stress tests and verification of dashboard accuracy.

• Ensure integrations adhere to standards such as GDPR, HIPAA, FedRAMP, and NIST AI Risk Management Framework (AI RMF) for secure and ethical AI operations.

• Bachelor's or Master's degree in Computer Science, Software Engineering, Data Science, or related discipline.
• Minimum 4+ years of experience in software engineering, AI integration, or cybersecurity, including production-level AI model deployment.

• Hands-on experience with observability and dashboard tools such as Grafana, Kibana, Prometheus, or Datadog.

• Familiarity with major cloud platforms (AWS, Azure, or Google Cloud) for AI model serving and orchestration.

• Proficiency in Python; additional experience in JavaScript, C++, or Go preferred.

• Experience with containerization and orchestration (Docker, Kubernetes) and API development (REST, GraphQL).

• Knowledge of logging frameworks (ELK Stack, OpenTelemetry) and visualization tools (Plotly, Chart.js).

• Understanding of AI model performance metrics (e.g., F1 score, precision, recall, latency) and drift detection methods (e.g., Population Stability Index, KS test).

• Knowledge of AI-specific vulnerabilities such as prompt injection, model inversion, and adversarial attacks, along with mitigation methods (e.g., differential privacy, model hardening, ART).

• Strong analytical and problem-solving capabilities for debugging complex integrations and optimizing performance.

• Effective communication skills to convey technical insights and system health metrics to technical and business audiences.

• Proven collaboration skills across multidisciplinary teams including Data Science, DevOps, and Cybersecurity.

• Must be U.S. Citizen and eligible to obtain a Department of Homeland Security (DHS) EOD clearance (requires a favorable background