IDB. AI for Financial Intelligence, Knowledge Graphs and Data Integration Intern, United States Washington D.C.

## IDB. AI for Financial Intelligence, Knowledge Graphs and Data Integration Intern, Washington D.C. - Organization:IDB - Inter-American Development Bank - Country:United States - Office:IDB Washington Posting End Date: 9/25/2026 11:59 PM EST The IDB Group is a community of diverse, versatile, and passionate people who come together on a journey to improve lives in Latin America and the Caribbean. Our people find purpose and do what they love in an inclusive, collaborative, agile, and rewarding environment. The IDB Group is looking for an active graduate student ound to be hired as an intern under the Consultant modality. The 2027 Winter Internship Program is a competitive paid opportunity, based in Headquarters, for university students from member countries to learn about development work and corporate management of the Bank, and to acquire work experience at the professional level through on-the-job learning. The Institutional Capacity of the State Division (ICS) is seeking a graduate intern to support its Transparency and Integrity thematic area. The intern will help develop and test artificial intelligence (AI) tools that make it easier to find, organize, and analyze information from large collections of government, administrative, and public records. Working alongside specialists in governance, data, and technology, the intern will support efforts to transform documents and databases into practical tools that help users identify connections, answer complex questions, and uncover patterns that would otherwise be difficult to detect. This includes helping to improve data quality, linking related information from different sources, and contributing to user-friendly applications that allow non-technical experts to access and use information more effectively. Specifically, the intern will support the design and prototyping of AI applications and interfaces that operate over large administrative and documentary databases. The work focuses on three connected areas: (i) retrieval-augmented generation (RAG) and graph-based retrieval (GraphRAG), combined with AI agents; (ii) AI-assisted extraction that turns digitized text and PDF collections into structured, query-ready datasets; and (iii) intelligent linking of related databases — including recovery from human data-entry errors in names, addresses, and geographic references — so that the resulting network of relationships can be analyzed and surfaced to users through well-designed interfaces. These capabilities will be applied to financial integrity use cases that are a priority for the region: financial intelligence analytics, anti-money laundering and countering the financing of terrorism (AML/CFT), traceability of financial transactions, and identification of ultimate beneficial owners across corporate, procurement, registry, and sanctions-related data sources. The intern will therefore see how AI methods translate into concrete tools that help supervisory authorities, financial intelligence units, and integrity teams detect risk patterns and follow the money across fragmented datasets. Prior knowledge of AML/CFT is not required — curiosity about the domain is. - Support the design of implementation blueprints and reference architectures for AI applications and interfaces that query, join, and present large structured and unstructured data sources used in financial intelligence and integrity analysis. - Prototype retrieval-augmented generation (RAG) and graph-based retrieval (GraphRAG) pipelines end to end — parsing, chunking, embedding, indexing, hybrid search, re-ranking, and citation of sources — over knowledge graphs of persons, companies, accounts, transactions, and jurisdictions, and measure retrieval quality against defined benchmarks. - Design, program, and evaluate AI agents and orchestrated multi-step workflows (tool use and function calling, guardrails, human-in-the-loop checkpoints, and automated evaluation harnesses), applying responsible-AI safeguards on privacy, explainability, and auditability that are appropriate for sensitive regulatory contexts. - Build pipelines that transform digitized documents (PDFs, corporate filings, registry extracts, free text) into structured databases, and support entity resolution and record linkage that reliably matches records affected by human errors in name spelling, transliteration, addresses, identifiers, and geocoding. - Explore AI-assisted network analysis over the linked databases — centrality, community detection, link prediction, and anomaly detection — to reveal ownership chains, control structures, and transaction paths, and document the work through architecture and data-flow diagrams, reproducible repositories, and presentations for technical and non-technical stakeholders. - Education: An active graduate student (master's or Ph.D.) in an engineering-based field preferred, including Computer Science, Software, Systems, Industrial or Electrical Engineering, Data Science, Artificial Intelligence, Applied Mathematics, Computational Statistics, or other related fields, from an accredited University. The graduation date should be after the Internship period. - Languages: Fluency in English with a working knowledge of one other Bank language (Spanish, French, or Portuguese) is a plus. - Key Skills: Learn continuously, collaborate and share knowledge, focus on clients, communicate and influence, innovate and try new things - Soft Skills: Result-driven, collaborative, innovative, client-oriented, strong communicator, self-motivated, eager to learn, and excellent time management skills. - Technical skills:Strong programming skills in Python and SQL; experience working with structured and unstructured data; familiarity with machine learning and large language model (LLM) applications; understanding of databases and data modeling; experience using APIs and Git-based version control; and the ability to clearly document technical work and communicate findings. Experience with AI application development, retrieval-augmented generation (RAG), vector search, knowledge graphs, document processing, or network analysis is desirable. Familiarity with tools such as LangChain, LlamaIndex, Elasticsearch, Neo4j, Streamlit, and related technologies is a plus, but candidates are not expected to have expertise in every tool listed.AI development environments and coding agents: including Claude (Claude Code, Cowork, Agent SDK, and the Model Context Protocol), OpenAI (API, Agents/Assistants, structured outputs), and comparable platforms.Agent and RAG frameworks: LangChain, LangGraph, LlamaIndex, Semantic Kernel or equivalents; evaluation and observability tooling such as Ragas, DeepEval, LangSmith, or Langfuse.Search and retrieval at scale: Elasticsearch or OpenSearch (BM25 and kNN hybrid search, aggregations, index design) and vector stores such as pgvector, FAISS, Qdrant, Weaviate, or Pinecone.Graph databases and network analysis: Neo4j and Cypher, Apache AGE or Amazon Neptune; NetworkX, igraph, or Graph-tool for network metrics; PyTorch Geometric or DGL for graph machine learning and link prediction; GraphRAG implementations (Microsoft GraphRAG, LlamaIndex property-graph index).Document AI and text-to-data: Azure AI Document Intelligence, AWS Textract, Google Document AI, Tesseract, Unstructured, Docling, or LayoutLM-family models; schema definition and validation with Pydantic or JSON Schema.Financial-integrity data and standards (desirable): familiarity with open corporate and beneficial-ownership data and registry datasets, and payment message standards such as ISO 20022; exposure to AML/CFT analytics or transaction-monitoring concepts is a plus, not a requirement.Interfaces and visualization: Streamlit, Gradio, or Dash for rapid prototypes; basic React/Next.js for application interfaces; Power BI, Plotly; and graph visualization with Cytoscape.js, Gephi, or similar. - AI development environments and coding agents: including Claude (Claude Code, Cowork, Agen