Efficiently Identifying Related Job Listings

Imagine you're working for a job board platform. A Product Manager (PM) is interested in implementing a "related jobs" feature on every job description page. This feature would display a list of jobs closely related to the one currently being viewed. You have some ideas on how to identify related jobs, such as using Natural Language Processing (NLP) techniques like bag of words and word embeddings. The PM specifies that "related jobs" are defined as those with similar job titles and descriptions. However, given the challenge, you realize that with millions of new job postings daily, identifying the top 10 related jobs for each one could be highly inefficient. Propose a system or method to effectively and efficiently identify the top 10 most related jobs for millions of new postings each day. Note: Assume there's an existing database with tens of millions of jobs that could potentially be related to each new job. Also, you have access to all job text features such as title, description, company, date, etc.