SC cleared - AI engineer

United Kingdom

Contract
£600 - £700 per day

Overview
A leading defence organisation is seeking an SC-cleared AI Engineer to develop fully air-gapped Retrieval-Augmented Generation (RAG) systems using open-source technologies. This is a rare opportunity to work on highly classified AI capabilities within secure, on-premises environments—no cloud, no external dependencies.
Key Responsibilities

• Architect and deploy end-to-end RAG pipelines using open-source LLMs (e.g. Llama 3, Mistral, Qwen) on isolated defence networks
• Build and manage local vector databases (Chroma, FAISS, Milvus) for sensitive document ingestion
• Host and optimise LLMs using vLLM or TGI on local GPU clusters (no internet access)
• Implement agent orchestration with LangChain/LangGraph in fully offline environments
• Design secure document processing workflows with robust data sanitisation protocols
• Develop monitoring and evaluation systems that operate within air-gapped infrastructure

Essential Requirements

• Active SC Clearance (mandatory) – DV clearance willingness is a plus
• Proven experience deploying open-source LLMs (Llama, Mistral, Falcon) in on-prem environments
• Expertise with local vector stores (Chroma, FAISS, Weaviate) in offline setups
• Strong hands-on experience with vLLM or Text Generation Inference for high-throughput serving
• Comfortable working in air-gapped systems with no access to external package repositories
• Experience with GPU orchestration (NVIDIA A100/H100) and CUDA optimisation
• Advanced Python skills, including offline dependency management and local mirrors

Technical Stack (All On-Premises)

• Models: Llama 3, Mistral, Qwen
• Vector Stores: Chroma, FAISS, Milvus
• Orchestration: LangChain, LangGraph
• Hosting: vLLM, TGI, Ollama on bare metal/private cloud
• Infrastructure: Air-gapped Kubernetes, local container registries

Desirable Skills

• Familiarity with defence/government IT security protocols
• Understanding of CIS benchmarks and NCSC guidelines
• Experience with cross-domain solutions and data diodes
• Knowledge of classification marking and secure handling procedures