
Lei Zhang from Bloomberg discusses the company's approach to AI in coding and software engineering. Bloomberg, with its large network, JavaScript codebase, and focus on the Bloomberg Terminal, has been exploring AI tools to boost productivity and stability. They started by deploying various AI solutions and measuring their impact, finding initial success in proof-of-concept and test writing. Bloomberg is focusing on evolving their codebase using uplift agents for patching and developing instance response agents to improve incident handling. To manage the deployment of these AI tools, Bloomberg has established a "paved path" with a gateway for model selection, a tool directory, and a platform for tool creation and deployment. They also incorporate AI coding into their training programs and leverage internal tech communities to promote adoption and knowledge sharing. Ultimately, Lei emphasizes that AI is changing the cost function of software engineering, prompting a reevaluation of what constitutes a high-quality software engineer.
Sign in to continue reading, translating and more.
Continue