AI/ML Engineer (Contract)

Aipex Defense; Northern (KY); Hybrid; USD 110,000 - 160,000; Full time

Job summary

AIPEX Defense is seeking a skilled engineer to build and validate embeddings and a retrieval-search pipeline for a US homeland-security project. You will stand up data-curation workflows and collaborate daily with a small distributed team, delivering on prototype, internal testing, and field-testing milestones.

The role emphasizes hands-on work with AWS Bedrock and Python, focusing on retrieval-based architecture rather than training from scratch.

Qualifications

• Hands-on experience with AWS Bedrock and Python.
• Comfortable building a retrieval-augmented system, not training from scratch.
• Able to work independently on a small, distributed team and communicate progress clearly.

Responsibilities

• Build and validate embeddings and the retrieval-search pipeline.
• Stand up the data-curation workflow, including synthetic data generation and validation.
• Support prototype deliverables and testing milestones tied to this pipeline.
• Document architecture and data-sourcing decisions for a non-technical reviewer.

Skills

Python

Retrieval-augmented systems

Independent work

Tools

AWS Bedrock

Job description

AIPEX Defense develops AI-enabled capabilities that support mission execution across defense and national security organizations. The company applies artificial intelligence, data integration, and modern digital technologies to improve decision-making, intelligence operations, and operational effectiveness in complex mission environments. We are small and early — a founding team, not a division of a holding company — and you should hear that from us rather than find it out. If you want your judgment visible in the finished thing, this is that job.

About AIPEX Defense

AIPEX Defense develops AI-enabled capabilities that support mission execution across defense and national security organizations. The company applies artificial intelligence, data integration, and modern digital technologies to improve decision-making, intelligence operations, and operational effectiveness in complex mission environments. We are small and early — a founding team, not a division of a holding company — and you should hear that from us rather than find it out. If you want your judgment visible in the finished thing, this is that job.

About the program

This role supports a prototype AI system that helps U.S. law enforcement and customs officials identify cultural property — art, antiquities, and heritage objects — that may be stolen or illegally trafficked. The architecture is retrieval-based: a query image is embedded and matched against a curated reference database of catalogued and at-risk cultural-property categories, rather than a classifier trained from scratch.

What you'll work on

• The core retrieval pipeline: AWS Bedrock multimodal embeddings, Bedrock Knowledge Bases, and OpenSearch Serverless similarity search against a curated reference corpus.
• A Claude-vision verification layer that reviews and explains candidate matches before they reach a human reviewer.
• The data-sourcing and curation pipeline for the reference corpus, built primarily from CC0/open-access collections (Smithsonian, the Met, Getty Open Content) and ICOM Red List categories, supplemented by AI-generated synthetic examples validated against real, held-out data.
• Close, day-to-day collaboration with the Application/QA Engineer on this same task order and with AIPEX's Program Manager/Technical Lead.

Responsibilities

• Build and validate the embeddings and retrieval-search pipeline described above.
• Stand up the data-curation workflow, including synthetic-data generation and validation against real examples.
• Support the Month 6 Prototype Application deliverable and the Month 7 Internal Testing and Month 8 Simulated Field Testing deliverables that depend on this pipeline.
• Document the architecture and data-sourcing decisions clearly enough for a non-technical Government reviewer to follow.

Qualifications AIPEX is looking for

• Hands-on experience with AWS Bedrock (or comparable embedding/retrieval infrastructure) and Python.
• Comfortable building and validating a retrieval-augmented system — deep classical computer-vision/classifier-training experience is not required, since this architecture is retrieval-based, not trained from scratch.
• Able to work independently on a small, distributed team and communicate progress clearly without daily in-person supervision.

Eligibility

This is a standalone, non-classified system that does not connect to any Department of State network and processes no non-public Government data. No security clearance, background investigation, or Common Access Card is required for this role. There is no U.S. citizenship requirement — you must be authorized to work in the United States, which is a standard requirement of any AIPEX engagement and not specific to this contract.