Intelligent recruitment agent for the financial sector
Nguyen, Sieng (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025112529651
https://urn.fi/URN:NBN:fi:amk-2025112529651
Tiivistelmä
This thesis is the design and assessment of an intelligent recruitment agent for the financial industry. This system is based on a serverless model and a chatbot, which is an OpenAI language model, as the point of entry for the applicants. It is possible to post a CV in several different formats, such as PDF, DOCX, or TXT format, or a LinkedIn profile. Each CV is read and scored by the system, and the applicants are provided with a tracking code whereby they are able to track their progress. This system is meant to facilitate a fair, transparent, and consistent hiring process besides minimising manual work among HR teams.
The prototype was built in phases and experimented with for the recruitment requirements in five generic financial positions. The solution relies on a Streamlit interface; the entire activity is stored in MongoDB Atlas and is based on the use of Google services to schedule e-mail and interviews. Jobs were allocated with established job requirements, and scenarios were set to test the system performance with respect to reaction to various applicant profiles.
It was tested that the system was capable of analysis of the CVs and scoring by application of rules and providing results in a trustworthy manner. It also minimised duplication of tasks, and it generated more understandable feedback to the candidates since it indicated the skills that influenced their mark. These characteristics contributed to ensuring that the assessment process was the same in all test cases.
To be used in actual recruitment, the system would be enhanced with more data and better integration with external HR systems. There is also the need to frequently verify the scoring rules and candidate feedback to provide fairness and accuracy in the dynamism of the market environment. Although speed and cost savings are achieved through automation, the HR people should still make the final hiring choices because it has to satisfy the ethical and regulatory demands in the financial industry.
The prototype was built in phases and experimented with for the recruitment requirements in five generic financial positions. The solution relies on a Streamlit interface; the entire activity is stored in MongoDB Atlas and is based on the use of Google services to schedule e-mail and interviews. Jobs were allocated with established job requirements, and scenarios were set to test the system performance with respect to reaction to various applicant profiles.
It was tested that the system was capable of analysis of the CVs and scoring by application of rules and providing results in a trustworthy manner. It also minimised duplication of tasks, and it generated more understandable feedback to the candidates since it indicated the skills that influenced their mark. These characteristics contributed to ensuring that the assessment process was the same in all test cases.
To be used in actual recruitment, the system would be enhanced with more data and better integration with external HR systems. There is also the need to frequently verify the scoring rules and candidate feedback to provide fairness and accuracy in the dynamism of the market environment. Although speed and cost savings are achieved through automation, the HR people should still make the final hiring choices because it has to satisfy the ethical and regulatory demands in the financial industry.