The Asian Infrastructure Investment Bank (AIIB) is a multilateral development bank whose mission is financing Infrastructure for Tomorrow in Asia and beyond—infrastructure with sustainability at its core. We began operations in Beijing in 2016 and have since grown to 111 approved members worldwide. We are capitalized at USD100 billion and AAA-rated by the major international credit rating agencies. Collaborating with partners, AIIB meets clients’ needs by unlocking new capital and investing in infrastructure that is green, technology-enabled, and promotes regional connectivity.

As AI models, including Large Language Models (LLMs) and agent-based generative AI, become increasingly embedded in financial workflows, ensuring robustness, reliability and explainability is critical. The next stage of AI adoption in risk management requires moving beyond experimentation toward deeper evaluation frameworks, model governance, and research-backed prototypes aligned with industry best practice and rating agency expectations.

This internship focuses on research and experimentation at the intersection of:

• LLM-based multi-agent systems for risk assessment workflows, especially validation, hallucination control, and structured decision reasoning.
• AI Model Risk concepts (e.g., alignment, robustness testing, reliability evaluation, benchmarking techniques, risk mitigation patterns).

The objective is to explore practical applications of these methods to AIIB’s risk use cases, such as early-stage risk detection, automated policy interpretation, structured expert review assistance, scenario-based risk reasoning and protocol of AI Model Risk governance.

The internship may result in:

• A research-quality paper in conference,
• technical research paper
• A prototype multi-agent workflow or evaluator framework, and
• A guideline research document supporting AI Model Risk guidance.

Responsibilities:

• A literature and benchmark review covering LLM multi-agent design, hallucination mitigation, alignment evaluation, and AI model risk practices in financial or regulated industries.
• Conceptual proposals demonstrating where multi-agent architectures can enhance risk decision processes.
• Development and testing of prototype(s) using one or more of the following:

prompt engineering frameworks, multi-agent orchestration frameworks, retrieval pipelines, or reinforcement learning-based evaluator loops.

• Evaluation of prototype robustness using metrics related to:

explainability, hallucination rate, consistency, repeatability, and governance requirements.

• A research paper, white paper, or equivalent publication-ready output summarizing findings.
• A brief methodology or playbook for potential adoption within AIIB.
• Prepare a transferable prototype or methodology guideline for future adoption within the risk management workflow.
• Present the completed project details in a slide deck / project document.

Requirements:

• Currently pursuing or recently completed a PhD in Machine Learning, Deep Learning, Generative AI, or a closely related field; Master’s candidates with strong research experience will be considered.
• Demonstrated research experience with publications in reputable ML/AI venues.
• Familiarity with LLM architectures, multi-agent designs, Retrieval-Augmented Generation, reinforcement learning, or model evaluation frameworks
• Demonstrated interest or experience in AI model governance, reliability testing, or alignment research.
• Proficiency in Python and modern LLM/agent frameworks (examples: LangChain, AutoGen, Hugging Face, OpenAI oSDK, DeepEval, RAG pipelines)
• Strong analytical and problem-solving skills, with the ability to handle complex, unstructured data sources.
• Fluent in English, with excellent written and verbal communication skills.
• Capable of working both independently and collaboratively, and willing to adapt to evolving project goals.

AIIB is committed to diversity, transparency, and inclusion. We believe our strength comes from having a team with the right diverse skills, experiences, and abilities selected through a merit-based competitive process. We actively encourage applications from people from both within and outside AIIB members, regardless of nationality, religion, gender, race, disability, or sexual orientation.

Join in AIIB’s mission to promote sustainable infrastructure investments and to improve social and economic outcomes in Asia and beyond.

Previous experience and qualifications will determine the grade and job title at which successful applicants will enter AIIB.