Assistant Research Scientist (PREP0004962

PREP Research Associate

Salary: $50,000-$70,000 a year

This position is part of the National Institute of Standards and Technology (NIST) Professional Research Experience Program (PREP). NIST recognizes that its research staff may want to collaborate with researchers at academic institutions on specific projects of mutual interest and, therefore, requires those institutions to be recipients of a PREP award. The PREP program involves the staff from a wide range of backgrounds conducting scientific research across various fields.

Individuals in this position will perform technical work supporting the collaboration's scientific research.

Citation Injector:
Source Attribution for AI-Generated Text

Overview: ITL's role in the Citation Injector project, "Citation Injector:
Source Attribution for AI-Generated Text," involves the following tasks: (1) segmenting AI-generated text into attribution units and retrieving candidate source chunks from a local corpus, and (2) validating and reranking candidate evidence to insert accurate, source-grounded citations and produce auditable quality metrics. Our success depends upon the availability of highly skilled domain experts. We are challenged with difficult tasks that require not only expertise in building agentic AI and LLM pipelines, but also in designing LLM-judge techniques for narrow entailment and support/refutation checks.

• Developing agentic AI evaluations and LLM pipelines to improve grounding, reliability, and traceability of AI-generated text.
• Building citation reconstruction pipelines that convert uncited AI outputs into source-grounded, cited answers.
• Designing LLM-judge techniques for narrow entailment and support/refutation checks to validate citation quality.
• Produce high-quality publications based on research and results present at internal and external meetings and conferences.

• US citizenship is preferred.
• An MSc degree in Computer Science or a related field, with experience in AI/LLM system development.
• Experience building and evaluating agentic AI systems and large language model (LLM) pipelines, including retrieval-augmented generation (RAG) and LLM-based judge/entailment evaluation.
• Ability to build deployable software solutions for AI-based source attribution and evaluation tasks.
• Strong oral and written communication skills and strong presentation skills.

Privacy Act Statement

Authority: 15 U.S.C. § 278g-1(e)(1) and (e)(3) and 15 U.S.C. § 272(b) and (c)

Purpose:

The National Institute for Standards and Technology (NIST) hosts the Professional Research Experience Program (PREP) which is designed to provide valuable laboratory experience and financial assistance to undergraduates, post-bachelor's degree holders, graduate students, master's degree holders, postdocs, and faculty.

PREP is a 5-year cooperative agreement between NIST laboratories and participating PREP Universities to establish a collaborative research relationship between NIST and U.S. institutions of higher education in the following disciplines including (but may not be limited to) biochemistry, biological sciences, chemistry, computer science, engineering, electronics, materials science, mathematics, nanoscale science, neutron science, physical science, physics, and statistics. This collection of information is needed to facilitate the administrative functions of the PREP Program.

NIST will use the information collected to perform the requisite reviews of the applications to determine eligibility, and to meet programmatic requirements. Disclosure of this information is also subject to all the published routine uses as identified in the Privacy Act System of Records Notices: NIST-1: NIST Associates.

Furnishing this information is voluntary. When you submit the form, you are indicating your voluntary consent for NIST to use of the information you submit for the purpose stated. By applying to a CHIPS-funded PREP opportunity, you also acknowledge that participation in the project requires signing a Non-Disclosure Agreement (NDA) prior to…