HR AI Agents - Automated Resume Evaluation System

This repository demonstrates three different approaches to building AI-powered HR resume evaluation systems using AWS services. All implementations provide comprehensive candidate assessment including resume parsing, skills analysis, gap identification, candidate rating, and interview preparation.

🏗️ Architecture Approaches

Step Functions Implementation (stepfunctions-agent/)

A workflow-based approach using AWS Step Functions to orchestrate multiple specialized Lambda functions, each handling a specific aspect of resume evaluation.

Bedrock Agent Implementation (bedrock-agent/)

A multi-agent AI approach using Amazon Bedrock Agents with collaborative AI agents that work together to provide comprehensive resume evaluations.

Strands Multi-Agent Implementation (strands-lambda-agent/)

A natural collaboration approach using Strands Agents SDK where specialized agents communicate in natural language to provide adaptive, comprehensive resume evaluations.

Strands AgentCore Implementation (strands-agentcore-agent/)

An enterprise-grade approach using Amazon Bedrock AgentCore Runtime to host Strands agents with enhanced security, observability, and scalability features.

🚀 Key Features

• Automated Resume Processing - Upload resumes to S3 and get automatic evaluations
• Comprehensive Analysis - Skills assessment, experience evaluation, and education review
• Gap Identification - Identifies missing skills and experience gaps
• Candidate Rating - Numerical scoring (1-5 scale) with detailed justification
• Interview Preparation - Generates structured interview questions and talking points
• Structured Data Storage - All results stored in DynamoDB with consistent schema

📊 Evaluation Components

All implementations provide:

1. Resume Parsing - Extract structured information from resume text
2. Skills Analysis - Identify technical and soft skills
3. Experience Assessment - Evaluate relevant work experience
4. Education Review - Assess educational background and certifications
5. Job Matching - Compare candidate profile against job requirements
6. Gap Analysis - Identify areas where candidate may need development
7. Candidate Rating - Numerical score with strengths and weaknesses
8. Interview Notes - Structured questions and areas to explore

🛠️ Technology Stack

Common Services

• Amazon Bedrock - AI/ML models for natural language processing
• AWS Lambda - Serverless compute for processing logic
• Amazon S3 - Document storage and event triggers
• Amazon DynamoDB - NoSQL database for evaluation results
• AWS IAM - Identity and access management

Step Functions Specific

• AWS Step Functions - Workflow orchestration
• Multiple Lambda Functions - Specialized processing steps

Bedrock Agent Specific

• Amazon Bedrock Agents - Multi-agent AI collaboration
• Agent Collaboration - Supervisor-collaborator pattern

Strands Agent Specific

• Strands Agents SDK - Natural language agent collaboration
• Multi-Agent Swarm - Dynamic agent coordination
• Custom Tools - Specialized HR evaluation tools
• 15-Minute Processing - Extended timeout for comprehensive evaluations
• Structured JSON Output - Consistent data format across implementations

📁 Project Structure

hr-agents/
├── README.md                    # This file
├── stepfunctions-agent/         # Step Functions implementation
│   ├── README.md               # Step Functions specific documentation
│   ├── template.yaml           # SAM template
│   ├── deploy.sh               # Automated deployment script
│   ├── functions/              # Lambda function code
│   └── statemachine/           # Step Functions definition
├── bedrock-agent/              # Bedrock Agent implementation
│   ├── README.md               # Bedrock Agent specific documentation
│   ├── template-agents-phase1.yaml  # Bedrock Agents template
│   ├── template-lambda.yaml    # Lambda functions template
│   ├── deploy.sh               # Automated deployment script
│   └── functions/              # Lambda function code
├── strands-lambda-agent/         # Strands Agent implementation
│   ├── README.md               # Strands Agent specific documentation
│   ├── template-infrastructure.yaml # Infrastructure resources (S3, DynamoDB)
│   ├── template-lambda.yaml    # Lambda functions and layer template
│   ├── deploy.sh               # Automated deployment script
│   ├── create_lambda_package.py # Lambda packaging with proper layer structure
│   └── functions/              # Lambda function code
├── strands-agentcore-agent/    # Strands AgentCore implementation
│   ├── README.md               # Strands AgentCore specific documentation
│   ├── template-infrastructure.yaml # Infrastructure resources
│   ├── hr_agent.py             # Main AgentCore agent code
│   ├── s3_processor.py         # S3 event processor Lambda
│   ├── deploy.sh               # Automated deployment script
│   └── requirements.txt        # Python dependencies
└── samples/                    # Sample data for testing
    ├── jobs/                   # Sample job descriptions
    └── resumes/                # Sample resumes

🚀 Quick Start

Step Functions Implementation

cd stepfunctions-agent
./deploy.sh
python upload_samples.py

Bedrock Agent Implementation

cd bedrock-agent
./deploy.sh

Strands Agent Implementation

cd strands-lambda-agent
./deploy.sh

Strands AgentCore Implementation

cd strands-agentcore-agent
./deploy.sh

📋 Prerequisites

• AWS CLI configured with appropriate permissions
• AWS SAM CLI installed
• Python 3.11 or later
• Access to Amazon Bedrock models in your AWS region

🔧 Usage

1. Deploy your chosen implementation
2. Upload job descriptions to the S3 bucket under jobs/ prefix
3. Upload resumes to the S3 bucket under resumes/ prefix
4. Monitor processing through CloudWatch logs
5. View results in the DynamoDB candidates table

📊 Sample Output

All four implementations produce structured evaluation data including:

• Candidate Information - Name, contact details, resume summary
• Skills Assessment - Programming languages, frameworks, tools
• Experience Analysis - Years of experience, relevant roles, achievements
• Education Evaluation - Degrees, institutions, alignment with role
• Rating & Recommendation - Numerical score and hiring recommendation
• Interview Preparation - Suggested questions and areas to explore

🔍 Monitoring & Debugging

• CloudWatch Logs - Detailed execution logs for all components
• DynamoDB Console - View processed candidate evaluations
• Step Functions Console - Visual workflow execution (Step Functions implementation)
• Bedrock Console - Agent invocation traces (Bedrock Agent implementation)

🛡️ Security Considerations

• Resume data contains PII - ensure proper encryption and access controls
• Use IAM roles with least-privilege access
• Enable CloudTrail for audit logging
• Consider VPC endpoints for private communication

💰 Cost Optimization

• Monitor Bedrock token usage and optimize prompts
• Use appropriate Lambda memory settings
• Consider reserved capacity for high-volume DynamoDB usage
• Implement lifecycle policies for S3 storage

🤝 Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Test thoroughly
5. Submit a pull request

📄 License

This project is licensed under the MIT License

---

Choose Your Implementation:

• Step Functions for complex workflows with detailed monitoring
• Bedrock Agents for AI-native multi-agent collaboration
• Strands Lambda for natural language agent collaboration with maximum flexibility
• Strands AgentCore for enterprise-grade agent hosting with enhanced security and observability

All four approaches provide identical functionality with different architectural patterns! Read my full article here