Witad Awards 2022: Data science professional of the year—Yimei Fan, Moody’s Analytics

Solving complex problems and learning from them is a process Yimei Fan enjoys. As director of software architecture at Moody’s Analytics, she is always looking for the next challenge and finding ways to bring efficiency to help people conduct business better.

In January 2021, she initiated a proof-of-concept (PoC) to automatically retrieve content related to a set of ESG metrics from unstructured data in PDF documents, webpages, and news.

After the success of that PoC, Fan planned the roadmap for a full-scale AI tool that can extract content for thousands of companies on hundreds of ESG metrics.

“I worked closely across internal teams to prioritize customer needs and plan the roadmap for an AI cognitive search tool. We transitioned from researching and initiating proofs-of-concepts to building a full-scale product. We adopted a fail-fast principle into the product development process, which quickly stabilized the product system and ensured its successful launch,” she says.

The AI tool enables Moody’s Analytics to measure and derive ESG scores for a large number of companies within a few months. Business units within Moody’s Analytics can incorporate ESG signals from the AI tool into Moody’s Analytics existing credit and compliance solutions so that clients get a more complete view of their portfolio risks. The tool can also be customized to extract information for other non-financial and financial domains that appeal to clients globally.

The tool uses deep learning models in computer vision for text and table extraction from unstructured data. “I stay on top of current trends in machine learning techniques by following the latest research, attending industry conferences and engaging with industry colleagues. I adopted the best suitable models and fine-tuned those models to specifically fit into ESG domains,” says Fan.

A mistake she says she has learned from is that early in her career as a data scientist, she focused on developing models with the best accuracy instead of putting herself in customers’ shoes and looking at the big picture. “I have learned that communication and collaboration with stakeholders and customers are critical in seeing the big picture and coming up with a solution that truly solves customers’ problems,” she adds.

Fan hopes to continue sharing her passion for machine learning with her peers and help women in the data science profession to reach their potential. “It’s important to me to network with like-minded women in the industry and trade stories and advice for success,” she says.