The Future of CICD Continuous Code Quality using AI

Slide 1

Slide 1 text

The Future of CI/CD: Continuous Code Quality using AI Bangalore Rajani Ekunde & Pratik Singh

Slide 2

Slide 2 text

Hi, I’m Rajani • DevOps Engineer (GlobalLogic) • Docker Captain • WTM Ambassador

Slide 3

Slide 3 text

Agenda • Understanding CI/CD • Code Quality Challenges in CI/CD Pipelines • Leveraging AI for Automated Code Analysis and Review in CI/CD • Workflow • Ollama & Gemini Approach • Demo • Challenges and Considerations

Slide 4

Slide 4 text

Understanding CI/CD • CI - Continuous Integration • CD - Continuous Deployment • Automates integration, testing, and deployment in development workflows. • Ensures faster delivery with reduced manual intervention and errors.

Slide 5

Slide 5 text

Code Quality Challenges in CI/CD Pipelines • Limited time for comprehensive manual code reviews. • Inconsistent adherence to coding standards across teams. • Delayed detection of critical security vulnerabilities and flaws. • Frequent code changes leading to integration conflicts. • Maintaining quality while meeting rapid deployment demands.

Slide 6

Slide 6 text

Leveraging AI for Automated Code Analysis and Review in CI/CD Workflows • AI quickly finds and fixes common coding mistakes. • Helps maintain coding standards across all projects. • Detects security vulnerabilities early in the process. • Saves time by reducing manual review efforts. • Provides suggestions to improve code performance.

Slide 7

Slide 7 text

Understanding the Workflow 1.Setup a set of Rules (RULES.MD) 2.API Key for AI Studio or Vertex AI 3.GPU CI/CD Runners 4.Basic Python 5.A can-do attitude!

Slide 8

Slide 8 text

1.Provision a GPU runner

Slide 9

Slide 9 text

2. Choosing the right LLM

Slide 10

Slide 10 text

3. Preparing the runner

Slide 11

Slide 11 text

4. Installing Ollama

Slide 12

Slide 12 text

5. Run the Python Script

Slide 13

Slide 13 text

Demo

Slide 14

Slide 14 text

Challenges and Considerations • AI may produce false positives, requiring manual verification. • Training AI models needs large, high-quality datasets. • Integration with existing CI/CD tools can be complex. • Ensuring AI adapts to evolving coding practices and trends. • Balancing automation with manual reviews for critical code areas.