Senior Principal Data Scientist / AI-ML SME (Analytic Superiority)

Leidos•Fort Meade, MD

7d•$154,050 - $278,475•Onsite

About The Position

The Leidos Intel Sector is looking for a premier AI/ML Subject Matter Expert (SME) to serve as a Technical Closer for our COSS 3.0 program supporting USCYBERCOM and the Cyber National Mission Force (CNMF) at Fort Meade, MD. In this elite role, you will architect and engineer the cross-platform AI frameworks required to achieve absolute analytic superiority. You will compress both defensive cyber operations (identifying network vulnerabilities and gaps) and offensive operations (vulnerability discovery and automated targeting) from days down to minutes. As a Technical Closer, you will mentor senior technologists by example—working "fingers-on-keyboard" to solve the command's most complex technical roadblocks and pushing production-grade AI directly into multi-cloud, hybrid, and air-gapped mission enclaves.

Requirements

15+ years of hands-on experience in software engineering, data science, or distributed systems.
5+ years of specialized experience in Machine Learning Engineering, deep learning, or LLM optimization.
3–5 years working within the DoD/IC cyber ecosystem, specifically building tools that map vulnerabilities or accelerate targeting cycles.
Active TS/SCI with Polygraph clearance.

Nice To Haves

Google Cloud Professional Machine Learning Engineer certification
AWS Certified Machine Learning – Specialty certification
NVIDIA Generative AI/LLM Associate certification
Prior certification as a CMF Exploitation Analyst (EA), Digital Network Analyst (DNA), or specialized technical experience as an Army 17A/170A, Navy 181X, or Air Force 17D/17S.

Responsibilities

Architect, deploy, and scale distributed AI workloads across any environment required, including AWS SageMaker, Google Vertex AI, Azure Government, and bare-metal, air-gapped server racks.
Deploy and optimize tools like LangGraph, CrewAI, or AutoGPT to automate cyber threat identification and offensive target generation at wire speed.
Fine-tune open-source large language models (e.g., Llama 3, Mistral) inside secure enclaves using PyTorch or TensorFlow.
Utilize NVIDIA TensorRT, Triton Inference Server, vLLM, and quantization libraries (bitsandbytes) to compress models for high-throughput execution under strict hardware constraints.
Direct the engineering of enterprise Retrieval-Augmented Generation (RAG) stacks using LangChain paired with high-performance vector databases like Milvus, Qdrant, or Pinecone.
Connect intelligent agents directly into security orchestration platforms (e.g., Palo Alto Cortex XSIAM/XSOAR) to trigger automated network defense actions and ingest massive, real-time PCAP and telemetry streams via Apache Kafka/Spark.
Demonstrate engineering mastery in Python, Go, Rust, and C/C++ to build ultra-fast cyber tools, write optimized GPU kernels, and interface with distributed frameworks like Ray.
Support both Defensive Cyber Operations (DCO) (log parsing, behavioral threat hunting, anomaly detection) and Offensive Cyber Operations (OCO) (automated vulnerability discovery, exploit generation, payload optimization).
Integrate custom AI/ML pipelines into unified mission systems and high-value data streams found across Project Maven, Palantir Foundry, and tactical command frameworks.