Teams at DEFCON AI

Build and own production entity-resolution capabilities, including blocking, candidate generation, pairwise scoring, clustering, threshold policies, deduplication, and known-entity checks. Develop provenance systems for traceable, explainable graph relationships and matching decisions. Evaluate matching performance, document implementation logic, and balance false merges against false splits. Work remotely with occasional travel while supporting government mission applications.

Extract structured facts from long, unstructured and semi-structured documents while preserving exact source citations. Validate evidence, distinguish stated totals from derived calculations, handle ambiguity without guessing, and apply established extraction methods consistently. Collaborate with entity resolution and retrieval teams to support downstream systems. The role requires strong Python and NLP or information-extraction experience, U.S. citizenship, and an active Secret clearance.

Build and operate production retrieval-augmented generation systems, including embeddings, vector retrieval, grounded language-model generation, citation validation, prompt and schema design, model packaging, serving, rollback, and telemetry. Develop bounded model assistance for narrative extraction with provenance and rule-context tracking. Maintain resilient serving paths in secure government cloud environments, including restricted or air-gapped deployments.