Research Internship Generative AI for Fire Modeling Research

Overview

FM Global is a leading property insurer of the world's largest businesses, providing more than one-third of FORTUNE 1000-size companies with engineering-based risk management and property insurance solutions. FM Global helps clients maintain continuity in their business operations by drawing upon state-of-the-art loss-prevention engineering and research; risk management skills and support services; tailored risk transfer capabilities; and superior financial strength. To do so, we rely on a dynamic, culturally diverse group of employees, working in more than 100 countries, in a variety of challenging roles.

Responsibilities

The Fire Modeling team is seeking a motivated and competent candidate as a summer intern or co-op student to leverage recent developments in Generative Artificial Intelligence (GenAI) and Large Language Models (LLMs) to develop a tool to greatly simplify the use of computational fire dynamics (CFD) software. This project will expand the use of CFD and its predictive capabilities to a wider audience. We anticipate the successful candidate will be an advanced undergraduate or graduate student in a technical field (e.g., computer science, data science,engineering, applied mathematics, etc.). The candidate need not be familiar with fire dynamics.

1. Under the supervision of senior scientists, configure an LLM to generate CFD simulation input files using a natural language interface, employing either an open-source LLM hosted locally, or a closed-source LLM via API. Depending on experience level, the candidate may choose to employ different approaches to configure or train the LLM to achieve the objective, including fine-tuning, few-shot prompting and other prompt engineering approaches.2. Develop a command-line application to agentize the LLM, generate the input files in-situ on a high-performance computing (HPC) system and initiate computation.3. Time permitting, further configure the LLM agent to perform basic post-processing of the data generated from the computation.4. Document the internship achievements and make technical presentations to scientists.

Qualifications

Currently enrolled and in a college or university and working minimally towards a bachelor degree BS degree in a technical field such as Electrical, Data Analytics, Mechanical, Chemical Engineering, Material Science or related fields.

Highly preferred graduate student or senior year undergraduate student. 

1. Proficiency in Python programming (e.g., numpy, scipy, pandas), manipulating data structures (e.g., json), and API interactions.2. Strong grasp of LLM manipulation through prompt engineering.3. Familiar with finding, modifying and utilizing open-source projects and repositories.4. Proficiency in using Linux systems and working on the command line (e.g., git, bash, vim).5. Good machine learning (ML) fundamentals (e.g., supervised vs. unsupervised learning).6. Foundational knowledge of natural language processing (NLP) and language models.7. Strong problem-solving and communication skills

FM Global is an Equal Opportunity Employer and is committed to attracting, developing, and retaining a diverse workforce.