Apply on Employer Site

BigBear.ai · 18 hours ago

AI/ML Security Integration Engineer

Annapolis Junction, MD

Full-time

Onsite

Senior Level, Lead/Staff

10+ years exp

BigBear.ai is a leading provider of AI-powered decision intelligence solutions for national security, supply chain management, and digital identity. They are seeking an AI/ML Security Integration Engineer to configure, fine-tune, and maintain an ATO Automation Platform’s Large Language Model capabilities to ensure secure operation within customer environments.

Artificial Intelligence (AI)GovernmentInformation TechnologyMachine LearningNational SecurityService Industry

No H1B

Security Clearance Required

U.S. Citizen Only

Responsibilities

Configure and optimize the ATO Automation Platform’s LLM backends (e.g. GPT-4, Claude Sonnet, Llama) based on customer security requirements and deployment constraints

Implement and maintain the Retrieval-Augmented Generation (RAG) architecture that powers the ATO Automation Platform’s automated compliance documentation

Develop and manage customer-specific knowledge bases incorporating organizational policies, security guides, and system documentation for AI processing

Fine-tune LLM parameters and prompt engineering to ensure outputs match organization-specific documentation standards and terminology

Implement secure data handling procedures for AI processing of potentially sensitive system configurations and security information

Configure vector databases and enterprise search technologies that support the ATO Automation Platform’s compliance knowledge graph

Monitor AI model performance and accuracy of generated compliance artifacts, implementing continuous improvement processes

Deploy approved LLM models in restricted environments including air-gapped networks and classified systems

Deploy and configure open-source Llama model in air-gapped DoD environment for processing IL5 system configurations

Fine-tune Claude Sonnet prompts to generate control implementation descriptions matching agency-specific SSP format requirements

Build custom RAG knowledge base incorporating customer’s 500+ security policies and standard operating procedures

Implement data sanitization pipelines to redact classified markings before AI processing

Optimize vector embedding strategies to improve the ATO Automation Platform’s accuracy in mapping code configurations to NIST 800-171 controls

Qualification

Large Language ModelsRetrieval-Augmented GenerationPythonVector databasesCloud-based AI servicesSecure AI deploymentPrompt engineeringAI governance frameworksNatural language processingMLOps practicesCybersecurity principlesAI accuracy assessmentCertifications