ML Engineer, Retrieval & Grounded Generation – Remote

DEFCON AI, Inc.; McLean (VA); À distance; USD 165 000 - 200 000; Plein temps

Avantages offerts par ce poste

Fully remote

Equity package

Health insurance

Unlimited PTO

Parental leave

Résumé du poste

Defcon AI in the United States is seeking an ML Engineer, Retrieval & Grounded Generation, to build embeddings, vector storage, and retrieval at scale, and to integrate language models with sources and citations tied to records in a secure government cloud environment.

You will own prompt design, model packaging, versioning, serving, and telemetry for measuring quality, with an emphasis on production-ready infrastructure rather than demos.

Qualifications

• 5+ years of production or near-production retrieval-augmented systems you built yourself.
• Detail-oriented discussion of retrieval design, vector store choice, and grounding tests.
• Strong Python, with hands-on embeddings and vector retrieval at scale.
• Clear statement of shipped work—prototype, proposal, or deployed code.
• US Citizenship and active US Secret clearance to start.

Responsabilités

• Implement embeddings, vector storage, and retrieval across a provenance-tracked base.
• Integrate language models with citations tied to source records and test for failures.
• Design prompts and output schemas; own packaging, versioning, serving, and rollback.
• Instrument telemetry for quality, attribution, grounding, latency, and throughput.
• Provide bounded model assistance with outputs tied to source passages.

Connaissances

RAG retrieval design

Python embeddings

Vector retrieval

US Citizenship

Secret clearance

Description du poste

Defcon AI in the United States is seeking an ML Engineer, Retrieval & Grounded Generation, to build embeddings, vector storage, and retrieval at scale, and to integrate language models with sources and citations tied to records in a secure government cloud environment.

You will own prompt design, model packaging, versioning, serving, and telemetry for measuring quality, with an emphasis on production-ready infrastructure rather than demos.