Data Scientist / AI Engineer

Phoenix Group of Virginia, Inc.; Norfolk (VA); On-site; USD 140,000 - 200,000; Full time

Job summary

Phoenix Group of Virginia, Inc. seeks a Senior Data Scientist / AI Engineer to design, deploy, and optimize production-grade RAG systems and agentic AI workflows for Carrier Team One (CT1) analytics.

You will turn complex naval maintenance data into actionable insights, automated summaries, and predictive signals to support decision making. Candidates should have deep LLM, MLOps, and cloud experience, strong Python skills, and the ability to translate Navy requirements into reliable AI solutions

Qualifications

• Experience with production-grade LLM/RAG systems and agentic workflows.
• Strong Python programming and data engineering skills.
• Ability to translate Navy requirements into reliable AI solutions.
• Experience with DoD/Navy data governance and RMF/ATO considerations.
• Proven ability to brief results to senior stakeholders.

Responsibilities

• Design, develop, deploy, and improve production-grade RAG systems and agentic AI workflows.
• Integrate data from NAVSEA, MAXIMO, Qlik, and other sources for analytics.
• Develop AI-enhanced features for dashboards and decision support with measurable ROI.
• Establish MLOps practices, versioning, monitoring, and testing.
• Collaborate with cross-functional teams and communicate complex concepts clearly.
• Maintain technical documentation and support audits and compliance.

Skills

LLM engineering

MLOps

Python

NLP

Stakeholder comms

Cloud deployment

Education

Master's degree or higher

Bachelor's degree in related field

Tools

LangChain/LangGraph

Databricks

AWS SageMaker

cURL

ADVANA

Job description

Position Summary

Carrier Team One (CT1) leads transformative data-driven process improvement and knowledge management initiatives for nuclear aircraft carrier maintenance and modernization. This includes maintenance support for all aircraft carrier availabilities, including Refueling Complex Overhaul (RCOH), Planned Incremental Availability (PIA), Selected Restrictive Availability (SRA), Docking Planned Incremental Availability (DPIA), and Carrier Incremental Availability (CIA). The Senior Data Scientist / AI Engineer serves as the principal technical expert and hands‑on builder for CT1's advanced analytics and artificial intelligence capabilities. This role directly supports the mission to accelerate capability delivery to the fleet, reclaim decades of obsolescence across alterations, and optimize the expenditure of thousands of man‑days and millions of dollars per availability. CT1 particularly values candidates who combine strong technical capability with exceptional learning agility, intellectual curiosity, and drive to rapidly master complex domains and deliver measurable results. This work builds upon CT1's existing Operations Dashboard and Qlik Sense analytics investments that have already demonstrated $9.3M+ savings and 5,200+ man‑day reductions per project.

Position Description

This is a senior technical individual‑contributor role (non-supervisory) requiring deep expertise in modern LLM engineering, production MLOps, and the ability to translate operational Navy requirements into reliable, secure, and measurable AI solutions. The position demands both exceptional technical craftsmanship and strong stakeholder communication skills to brief results and limitations to project superintendents, engineers, senior leadership, and external stakeholders (NAVSEA, PEO, TYCOM).

Primary Functions

The primary focus of this position is the design, development, deployment, and continuous improvement of production‑grade Retrieval-Augmented Generation (RAG) systems, agentic AI workflows, and LLM-powered analytics tailored to carrier hotwashes (after‑action reviews), project performance data, technical documentation, and operational decision support. The incumbent will leverage and extend platforms including ADVANA, Databricks, AWS SageMaker, LangChain/LangGraph, and cURL to turn complex, unstructured, and structured naval maintenance data into actionable insights, automated summaries, predictive signals, and natural‑language query capabilities.

Position Requirements

35% — LLM/RAG Pipeline Architecture, Development & Productionization

Lead the end-to-end design and implementation of scalable, secure RAG and multi-agent systems. Select and optimize embedding models, chunking strategies, hybrid retrieval (vector + keyword + metadata), reranking, and context compression techniques specifically tuned for technical naval maintenance documentation, alteration history, hotwash narratives, and project artifacts. Implement hallucination detection, citation/grounding mechanisms, and domain-adapted evaluation metrics. Deploy and iterate on LangChain/LangGraph (or equivalent) orchestration layers integrated with approved LLM endpoints (Gemini primary; others as authorized). Ensure solutions meet performance, cost, latency, and reliability targets for operational use.

Architect and maintain robust data pipelines that ingest, clean, enrich, version, and serve data from heterogeneous shipyard sources (ADVANA datasets, MAXIMO or equivalent EAM systems, Qlik extracts, project schedules, technical manuals, and unstructured logs). Implement vector stores, metadata filtering, and feature stores on Databricks or SageMaker. Establish data quality monitoring, lineage, and governance aligned with DoD and Navy data standards. Enable both batch and near-real-time capabilities as required for hotwash cycles and availability execution.

15% — AI‑Augmented Operational Analytics & Decision Support

Partner with CT1 analysts, Work Integration Managers, Assistant Project Superintendents, and Process Masters to identify high-impact AI use cases. Build and productionize AI‑enhanced features for the CT1 Operations Dashboard and related tools: natural language querying of project data, automated hotwash summarization and insight extraction, risk flagging, duration/resource forecasting, and semantic search over historical alterations and lessons learned. Quantify and communicate ROI in terms of man‑days saved, cost avoidance, schedule compression, and improved decision quality.

10% — MLOps, Evaluation, Monitoring & Responsible AI

Establish production MLOps practices: prompt/model versioning, CI/CD for pipelines and agents, automated regression testing, drift detection, cost tracking, and observability. Design and maintain rigorous, Navy-context-specific evaluation harnesses (offline benchmarks + online A/B or human feedback loops). Champion and implement DoD AI ethical principles, bias auditing, transparency, human-in-the-loop safeguards, and compliance with emerging Navy/DoD AI governance and cybersecurity requirements (including RMF/ATO considerations for any new capabilities).

10% — Cross‑Functional Collaboration, Validation & Knowledge Transfer

Serve as the primary AI technical liaison to CT1's cross-functional teams and the broader Knowledge Management Community of Practice (KM COP). Conduct requirements workshops, demo iterations, and validation sessions with subject‑matter experts (welders, planners, engineers, logisticians). Translate complex technical concepts and model limitations into plain language for senior decision‑makers. Document architectures, runbooks, prompt libraries, and lessons learned. Actively contribute to CT1's knowledge management, process improvement, and innovation initiatives, including agentic research efforts.

Continuously scan the rapidly evolving LLM/agent/RAG landscape for high-value, low-risk capabilities that can be adopted within approved cloud and security boundaries. Rapidly prototype promising approaches against real CT1 use cases (e.g., multi‑agent hotwash analysis, knowledge graph augmentation of RAG, predictive signals from unstructured maintenance text). Provide concise technology assessments and recommendations to leadership.

Mentor junior data professionals, contractors, or rotating personnel on best practices. Maintain living technical documentation and contribute to CT1's knowledge base. Support audits, data calls, and continuous monitoring requirements. Identify process or tooling improvements that increase team velocity and solution quality. Perform other related duties as assigned in support of CT1 mission objectives.

Required Technical Competencies

• Expert-level knowledge and hands-on production experience with modern LLM engineering, RAG architectures, agentic workflows (LangGraph or strong equivalent), prompt engineering, evaluation frameworks, and grounding/citation techniques.
• Advanced proficiency in Python and the LLM/data ecosystem: LangChain/LangGraph (or LlamaIndex + custom orchestration), vector databases, embedding models, Hugging Face Transformers (as needed), Pandas/Polars, SQL, and Spark/Databricks Delta Lake.
• Strong practical experience deploying and operating ML/AI workloads on cloud platforms, with preference for AWS SageMaker and/or Databricks; equivalent experience on Azure ML or Google Vertex AI is highly transferable.
• Demonstrated ability to build production data pipelines, implement MLOps (CI/CD, monitoring, versioning), and manage the full lifecycle of AI solutions from prototype through sustained operations with measurable SLAs.
• Solid understanding of NLP techniques for technical and semi-structured text (chunking, entity extraction, summarization, semantic search) and experience applying them to real-world operational or maintenance datasets.
• Ability to rapidly acquire and apply context from complex naval maintenance, engineering, logistics, and project management domains; prior DoD/Navy/shipyard or heavy industrial experience is a strong plus but not mandatory if accompanied by proven ability to learn technical domains quickly.
• Excellent written and oral communication skills, including the ability to produce clear technical documentation and to brief technical and non-technical audiences up to senior executive/flag level on capabilities, trade-offs, risks, and measured outcomes.
• Strong collaboration and facilitation skills; comfortable leading requirements workshops, validation sessions, and iterative co-design with domain experts who may have limited AI background.
• High degree of self-motivation, intellectual curiosity, and disciplined execution in a fast-paced operational environment with competing priorities and evolving requirements.

Preferred / Highly Desirable

• Active or recent Secret (or higher) security clearance.
• Prior experience supporting Navy, NAVSEA, shipyard, or other DoD maintenance/modernization analytics or AI initiatives.
• Hands‑on familiarity with ADVANA, Databricks Unity Catalog, or Navy/DoD data platforms and governance frameworks.
• Experience with knowledge graphs, hybrid search, multi-modal models, or LLM fine-tuning (parameter-efficient or continued pre-training) in regulated environments.
• AWS Certified Machine Learning – Specialty or equivalent cloud ML certification; relevant LLMOps or MLOps certifications.
• Track record of shipping production AI features that delivered quantified operational or business impact in complex environments.

Education

Master's degree or higher from an accredited institution in Data Science, Computer Science, Artificial Intelligence, Machine Learning, Statistics, Operations Research, or a closely related quantitative field is strongly preferred. A Ph.D. is advantageous for roles with significant research/prototyping elements but is not required. Or a Bachelor's degree in the same fields, combined with strong demonstrated impact on complex LLM/RAG or ML systems plus exceptional learning agility may be qualifying.

Experience

Generally 4–6+ years of professional experience, with stronger emphasis on independent ownership of production or near‑production RAG/agentic LLM systems, deeper technical leadership, and the ability to operate with minimal supervision on complex, high-stakes problems from day one.

Or Generally 2–4 years of professional experience in data science, machine learning engineering, or AI application development. Candidates must demonstrate clear, meaningful contribution to LLM, RAG, or other complex ML/AI systems (production, near‑production, or high-impact pilot systems that delivered measurable value). Exceptional learning agility, intellectual curiosity, and drive are heavily weighted. Outstanding portfolios or rapid progression on complex technical projects can offset modestly lower years of experience.

All candidates must show, through resume, projects, and interview, meaningful personal contribution to the design, implementation, significant improvement, or successful adoption of RAG, agentic LLM, or other complex ML/AI systems applied to technical or operational use cases. Evidence of rapid learning, high-quality delivery under ambiguity, intellectual curiosity, and measurable impact will be weighted heavily.

Purely academic, notebook‑only, or low-impact proof-of-concept work without clear stakeholder value or learning agility will generally not meet this factor at either level.

Work Environment and Physical Requirements

• Security Clearance — Secret clearance.
• Telework / Hybrid — Regular telework or hybrid arrangement (typically 2–3 days per week on-site or as mission dictates) is available and encouraged where duties permit. Some work (classified discussions, certain data access, collaboration sessions, shipyard walkthroughs) will require on-site presence at naval facilities.
• Travel — TDY (estimated 10 - 12 trips per year) to naval shipyards, or conferences for coordination, requirements gathering, training, or knowledge sharing.
• Cybersecurity & AI Governance — Must comply with all applicable DoD, Navy; cybersecurity policies, AI use guidelines, data handling requirements (including CUI and classified information), and RMF/ATO processes for any new capabilities developed or integrated.
• Ethics & Standards — Incumbent is expected to model the highest standards of professional conduct, intellectual honesty, and commitment to responsible, mission-aligned AI development.