Health eCareers: Using Generative AI to reinvent the Job Search Experience

Опубликовано 4 июня 2024, 18:10

The technical architecture behind this innovative job search tool is both robust and efficient, tailored to meet the specific needs of healthcare professionals. Initially, a user enters their query in natural language through the Health eCareers website’s interface. This input is processed by .NET application, which leverages Amazon Bedrock SDK hosted on ECS (Elastic Container Service). The prompt undergoes sanitation and is enhanced by Anthropic Claude via AWS Bedrock to detect and interpret geographic references, turning phrases like “Southern California” into actionable geo-coordinates for radius searching. Once refined, the prompt is vectorized using Amazon Titan and then sent to ElasticSearch hosted on EC2, which performs a vector search across our comprehensive job posting repository—continuously updated and vectorized using  Amazon Titan every six hours from various feeds. Following this, the top job postings are selected and used in conjunction with the original query in a  process facilitated again by Amazon Bedrock SDK in our application. This process utilizes the generative capabilities of Anthropic Claude to produce tailored responses, which are then presented to the user alongside the job postings. This system not only provides a seamless chat experience—allowing users to delve deeper into the job details or restart their search—but also significantly simplifies the job hunting process.

Check out more resources for architecting in the #AWS​​​ cloud:
amzn.to/2ZIbygO

#AWS #AmazonWebServices #CloudComputing #ThisIsMyArchitecture #ECS, #AmazonTitan, #AmazonBedrock, #AmazonS3, #EC2