Senior AI / Data Architect (Army Vantage & LLM Infrastructure)Location: Onsite 5 days per week at Picatinny Arsenal, Dover, NJ (Remote work is not permitted)
Clearance: U.S. Citizenship REQUIRED; Ability to obtain and maintain a DoD Security Clearance
Experience Level: Senior / Lead Level
Role OverviewWisEngineering is seeking a
Senior AI / Data Architect to lead the design, deployment, and optimization of end-to-end AI pipelines and data architectures supporting Department of Defense (DoD) programs at Picatinny Arsenal.
This is a high-impact, hands-on architectural role. You will be responsible for bridging enterprise defense data platforms like
Army Vantage (Palantir) with localized Large Language Model (LLM) infrastructure. The ideal candidate has direct experience building and deploying end-to-end AI pipelines, orchestrating unstructured data ingestion, and managing local LLM environments within secure DoD ecosystems.
Key Core ResponsibilitiesAI Pipeline & LLM Architecture- End-to-End AI Pipeline Deployment: Architect, build, and maintain production-grade AI/LLM pipelines connecting raw, unstructured data sources directly to localized models and analytics platforms.
- Palantir Vantage Integration: Design and deploy AI-enabled workflows, data ontologies, and integrated applications within Army Vantage (Palantir) to visualize real-time operational intelligence.
- Local Infrastructure Management: Serve as the Primary Technical Lead for locally hosted LLM environments (utilizing Ollama, LM Studio, and the Gemma model family). Perform complex system troubleshooting, parameter tuning, and performance optimization.
- Advanced Prompt Engineering & Fine-Tuning: Develop, test, and refine deterministic prompts and retrieval-augmented generation (RAG) frameworks to extract summaries, classifications, and actionable research from local LLMs.
Data Engineering & Scraping- Targeted Web Scraping & Ingestion: Write and maintain robust, safe Python web-scraping and extraction scripts to harvest, clean, and structure external and internal data needed to ground local LLMs.
- Unstructured Data Transformation: Build automated pipelines to process raw, unstructured formats (PDFs, web pages, APIs, free-text) into structured vector stores and data models.
- Security & DoD Compliance: Ensure all data pipelines, web harvesting, and LLM querying strictly adhere to DoD security, governance, and compliance guidelines.
Business Intelligence & System Deliverables- Enterprise Dashboards: Design, build, and maintain high-impact executive performance metrics and interactive dashboards using Microsoft Power BI and DAX.
- Data Tracking & Automation: Integrate Microsoft Platform environments (SharePoint, Microsoft Lists) with Python automation to drive automated reporting.
- Executive Deliverables: Generate automated intelligence reports, documentation, and executive deliverables in PDF, Word, and HTML using open-source Python libraries.
Required Qualifications & Technical Stack- U.S. Citizenship: Required due to DoD site access and security clearance requirements.
- Palantir / Vantage Expertise: Proven hands-on experience building pipelines, workflows, or applications within Palantir Foundry / Army Vantage.
- AI & LLM Frameworks: Demonstrated experience building end-to-end AI pipelines, RAG architectures, and managing local open-source LLM environments (Ollama, LM Studio, LangChain/LlamaIndex, Gemma).
- Core Languages: Advanced proficiency in Python 3 (Pandas, NumPy, BeautifulSoup/Scrapy, open-source document parsing) and SQL.
- Business Intelligence: Advanced Power BI modeling (DAX, complex relationships) and Microsoft Power Platform integration.
- Education & Experience: Bachelor's or Master's degree in Computer Science, Data Engineering, Artificial Intelligence, or a related technical field, paired with 5+ years of progressive technical experience as a Data/AI Architect or Senior Engineer.
Preferred Qualifications- Prior experience working onsite with Army DEVCOM, DoD, or military contractor environments.
- Active Secret (or higher) DoD Security Clearance Preferred.
