Applied AI Engineer Intern

Conduent State Healthcare, LLC•, Remote US

21h•$43,000 - $54,000•Remote

About The Position

We are seeking a remote Applied AI Engineer Intern to support AI-enabled initiatives within Conduent Government Healthcare Solutions, focused on organizing enterprise knowledge, improving information retrieval, and helping teams make better use of existing data and documentation. This internship is remote and is expected to work 20 hours per week. This posting represents a focused internship opportunity for a student interested in applying AI engineering, data analysis, automation, and emerging technology concepts to practical business challenges. The intern will gain hands-on experience supporting applied AI projects that may include reviewing and organizing enterprise content, assisting with data preparation and information-extraction workflows, evaluating retrieval approaches, supporting AI-enabled knowledge access, and documenting findings, limitations, and recommended next steps. This internship is part of a structured fall program that includes mentorship, networking opportunities, and exposure to senior leadership.

Requirements

Currently enrolled in a graduate program, or have equivalent preparation, in Data Science, Computer Science, Natural Language Processing, Information Retrieval, Machine Learning, Artificial Intelligence, Engineering, or a closely related field.
Coursework, project experience, or demonstrated work involving natural language processing, information extraction, vector search, retrieval-augmented generation as an experimental method, knowledge representation, applied AI for unstructured enterprise text, or experimental design and evaluation.
Comfort working with messy, unstructured text and conducting quantitative evaluation and human review—not only developing demonstrations.
Ability and willingness to build and empirically compare multiple representations rather than relying on a single vector index.
Strong written communication skills and the ability to explain technical work clearly to both technical and non-technical audiences.
Ability to work independently in a remote environment while collaborating effectively, incorporating feedback, and providing regular progress updates.

Nice To Haves

Familiarity with cloud platforms, particularly Microsoft Azure, is helpful; approved project resources will be provided by Conduent.

Responsibilities

Work with a team-provided, sampled, design-only corpus to assess document types, metadata, domain concepts, relationships, and quality issues that affect retrieval; scope will be established with the team.
Build preprocessing and LLM-assisted extraction workflows that transform unstructured design artifacts into candidate structured representations, including concepts, relationships, provenance, and concise summaries, with human validation.
Establish an embeddings/vector-index retrieval baseline as an experiment on the sampled corpus and compare it with richer approaches; the baseline is not intended to serve as a production tester-facing tool or test-authoring pipeline.
Explore knowledge graphs, knowledge maps, and other structured indexes, and use evidence to compare their performance with the vector baseline without assuming that one representation will be superior.
Use sequential LLM-assisted distillation and assimilation to produce a reusable written knowledge digest, or equivalent, layered on top of the index and/or graph for a representative sample selected with the team.
Demonstrate, with supervisor review, that retrieved and distilled knowledge is grounded and reusable through identifiers, provenance, and human-checkable excerpts.
Define and track retrieval and knowledge-quality measures, such as relevance, coverage, precision of retrieved context, redundancy, provenance and groundedness, completeness, consistency, and human validation, including review with QA practitioners.
Document architecture decisions, experiments, results, limitations, failure modes, and recommended next steps, including how Conduent teams can maintain the Azure-based experiments.