Numino Recruitment Process Automation Using GenerativeAI

Overview

At Numino Labs, the hiring process for Lead Technology Developer roles, involves a thorough review of candidate resumes. The traditionally employed keyword-based rating system to screen resumes presented significant limitations, leading us to integrate OpenAI’s advanced natural language processing capabilities into our resume rating process. This case study outlines our transition from a keyword-based system to an AI-driven evaluation method, detailing the implementation process, challenges, and outcomes.

As of June 2024, the AI resume rating has been operational in production for over 3 months, serving 3 critical positions

  • An average of 200 resumes per month per position
  • Accuracy rate of over 90%.
  • Average cost of 30$ / month towards Open ai.
Project Info

Category

AI, Cloud Backend

Client

Numino

Tags

  • AWS
  • OpenAI
  • Serverless

Product Features

Configurable Role-Based Prompts Management using Google Sheets
Workflow Orchestration using AWS Step Functions
Asynchronous communication between microservices using AWS SNS/SQS
Result Storage and Accessibility
Cost and Rate Limiting Management

Shortcoming of traditional keyword-based rating system

Contextual Irrelevance
Contextual Irrelevance
Keywords in resumes were often found in irrelevant sections, such as the skills section, without actual experience backing them up
Lack of Synonym Recognition
Lack of Synonym Recognition
Candidates described their experiences using different terminologies, resulting in missed qualifications due to the system’s inability to recognize synonyms or related terms.

OpenAI-Powered resume evaluation

Approach
To overcome the limitations with traditional keyword based rating, we defined a comprehensive set of criteria that could be more accurately assessed by an AI system
Workflow

The workflow consists of the following steps –

Architecture

The architecture of our AI-enhanced resume rating system leverages several technologies to create a seamless, efficient, and scalable solution. Here’s an overview of the components and their interactions

Role-Based Prompts Management using Google Sheets
Role-Based Prompts Management using Google Sheets
  • Purpose: To store role-specific prompts that guide the AI in evaluating resumes.
  • Benefit: Easy access and update capability for HR personnel.
  • Details: Each row in the Google Sheet represents a specific role and contains the tailored prompts for that role.
Workflow Orchestration using AWS Step Functions
Workflow Orchestration using AWS Step Functions
  • Purpose: To orchestrate the workflow of the resume evaluation process.
  • Benefit: Easy access and update capability for HR personnel.
  • Details: AWS Step Functions manage the entire process from receiving the resume, invoking Lambda functions, to writing the outcome back to Google Sheets.
  • Processing Logic included AWS Lambda - a serverless architecture that scales automatically with demand and reduces operational overhead & handles the interaction with OpenAI’s API.
Cost and Rate Limiting Management
Cost and Rate Limiting Management
  • Resume Size Check to control the cost associated with processing resumes with large sizes. Resumes exceeding a predefined size are flagged for manual review.
  • Serialization to handle OpenAI’s token per minute (TPM) rate limit. Ensures compliance with OpenAI’s usage policies and prevents request throttling
Result Storage and Accessibility
Result Storage and Accessibility
  • Results of the resume evaluations written back into a Google Sheet
  • Provides HR with a centralized and easily accessible location to review the AI-generated ratings and assessments.



Performance Metrics

Implementation of the AI-enhanced resume rating system consisted of several rounds of testing to ensure its accuracy and reliability

  • Pilot Phase: started with a pilot phase where we ran a subset of resumes through the new system and compared the results with the manual verification.
  • Feedback Loop: We gathered feedback from hiring managers and HR personnel to refine the prompts and improve the accuracy of the AI’s assessments.
  • Full Deployment: After successful testing and feedback incorporation, we fully deployed the AI-enhanced system into our hiring process.
Below shows a concise summary of the iterative testing and refinement of the prompt to achieve high accuracy and reliability.

Result / Summary

The above describes the state of the solution in production. By integrating OpenAI into our resume rating process, we have achieved the following results

Significantly improved the accuracy and efficiency of candidate evaluations that allowed us to better understand the true qualifications and experiences of candidates, leading to more informed hiring decisions and a more efficient hiring process overall.

Leveraging AWS infrastructure and OpenAI's capabilities provides for a robust, scalable, and efficient solution.

Read More

Pixm Anti Phishing v2.0
Cloud Backend,Web Apps,
Pixm Anti Phishing v2.0

Pixm is an AI tool that stops phishing attacks instantly when clicked, using serverless technology for scalable, cloud-based threat detection and comprehensive enterprise security management.
StrategicIT ePMOLite Project Management
Cloud Backend,Web Apps,
StrategicIT ePMOLite Project Management

ePMOLite transforms IT PMO for small to mid-sized organizations with a cloud-based, serverless solution that tracks progress and financials, facilitating informed decision-making.
PromptMule Prompt Cache LLMs
AI,Cloud Backend,
PromptMule Prompt Cache LLMs

PromptMule, a managed caching service, leverages semantic caching and vector embeddings to deliver lightning-fast responses, reducing costs by 70% while improving application responsiveness by 5x.