AI Engineer (Python/LLM/Data Parsing) - 50% telework

If you've shipped real AI systems, not just demos, and you want to build them on hardware you can actually touch, keep reading.

Linchpin Software is a small, growing company, and we're hiring AI Engineers in Chantilly, VA (50% telework). You'll build production AI that runs entirely on-prem: locally hosted models on GPU servers, in air-gapped and cross-domain environments, with no calls out to a public API. The work is hands-on engineering: parsing messy documents and data, extracting and resolving entities, building retrieval pipelines, and turning model output into structured, traceable results that analysts can trust.

We'll be honest: this is a new and growing body of work, so you'll help decide how it gets built.

Here's what we care about, and we hire for it:
- Be helpful. Answer the question, share what you know, and leave things better than you found them.
- Take care of the customer. Our customers have hard missions. We listen, follow through, and make their day easier.
- Be a good teammate. Review the code, jump in on the tough bug, cover for each other, and celebrate the wins together.

We also build software differently. Our engineers use AI-assisted development tools (Claude Code, Codex, GitLab Duo) every day to design, code, test, and debug faster.

What You'll Do:
- Build production AI/ML and GenAI pipelines, models, and microservices in Python
- Build LLM applications: embeddings, vector databases, retrieval, agentic workflows, and tool calling
- Engineer complex data parsing and ingestion for diverse structured and unstructured data
- Implement named entity recognition, entity extraction, and entity resolution
- Deliver multi-step retrieval and structured output with deep source citation
- Train, evaluate, and serve models (PyTorch, TensorFlow, scikit-learn) with MLOps and CI/CD
- Deploy on Kubernetes and locally hosted GPU servers, and tune for accuracy, latency, and cost

Required:
- Active Top Secret, TS/SCI, or Intel Agency clearance
- Bachelor's degree
- 10+ years of engineering experience with substantial hands-on AI/ML work
- Advanced Python, with strong algorithms, data structures, and distributed systems fundamentals
- Hands-on experience deploying production GenAI/LLM systems, embeddings, and vector databases
- PyTorch, TensorFlow, or scikit-learn
- MLOps tooling (MLflow or similar), containers, and Kubernetes
- Background in NLP, computer vision, or a related AI field
- Clear communication and a genuine willingness to help customers and teammates

Preferred:
- Self-hosting open models on GPUs (vLLM, Ollama, Hugging Face, or similar)
- Document parsing and extraction at scale
- Vector and graph databases (LanceDB, pgvector, Neo4j, or similar)
- FastAPI, PostgreSQL, Temporal.io
- Geospatial, temporal, or computer vision AI
- Air-gapped, multi-enclave, or cross-domain environments
- AI governance and secure development practices

The Honest Details:
- Location: Chantilly, VA, 50% telework. You'll be on site about half the time, and we mean it.
- Pay: $130,000 - $230,000, based on experience and what you bring.
- Clearance: active Top Secret, TS/SCI, or Intel Agency clearance required. We can't sponsor one for this role.

What You'll Gain:
- Real, production AI engineering on a mission that matters
- A chance to shape a new program area from the ground up
- A small team where your work is seen and your opinion counts
- Competitive compensation and benefits

If you like building AI systems that actually work in the real world, apply. A real person on our team reads every application.