Tutorial B

Industrial DevOps Workshop with LLM/AI Integration

Fees

Date

Time

Overview

Module 1: Introduction to Agile, DevSecOps, and AI in Safety-Critical Environments

• • Overview of Agile principles, DevSecOps culture, and their synergy with AI tools
  • Challenges and opportunities in integrating AI/LLMs into safety-critical cyber-physical systemS
  • Case studies of Agile/DevSecOps and AI implementation opportunities in large-scale, safety-critical industries
  • Ethics and governance in AI-assisted development workflows

Module 2: Adapting Agile Practices for Safety-Critical Development with AI/LLMs

• • Agile methodologies (Scrum, Kanban, SAFe) and scalability for safety-critical systems with AI adoption
  • Using AI/LLMs to enhance backlog management, task prioritization, and user story generation
  • Tailoring Agile to comply with safety standards while leveraging AI tools
  • Risk management and iterative development with AI/LLM-driven risk assessment

Network Break

Module 3: Integrating Security, Compliance, and AI into the Development Life Cycle

• • Understanding DevSecOps principles
  • How AI enhances threat modeling, vulnerability detection, and automated risk mitigation
  • Addressing AI-specific cybersecurity challenges (e.g., prompt injection attacks, model poisoning risks)

Module 4: Engineering for Dependability, Resilience, and AI Verification

• • Ensuring resilience and fault tolerance
  • Automated testing strategies for safety-critical functions
  • Design principles for safety-critical cyber-physical systems with integrated AI components
  • AI/ML model verification and validation to ensure reliability in safety-critical functions

Lunch Break

Module 5: Continuous Integration and Continuous Deployment (CI/CD) with AI in Regulated Environments

• • Balancing speed, precision, and safety in continuous deployment with AI accelerators
  • Setting up CI/CD pipelines; considerations for AI-driven tools for code review and error prediction
  • Using AI to automate compliance verification in CI/CD pipelines
  • Real-world challenges in deploying AI/LLM-powered applications at scale

Module 6: Systems Thinking, Complexity, and AI-Driven Design

• • Applying systems thinking to cyber-physical systems augmented by AI/LLMs
  • Simulation, modeling, and formal methods
  • Managing interdependencies and complex architectures with AI-powered modeling and simulation tools
  • Addressing emergent issues introduced by the use of AI models in the system lifecycle

Network Break

Module 7: Collaboration and Communication with AI in DevSecOps

• • Leveraging AI/LLM tools to enhance cross-functional team collaboration
    • Example: AI-driven documentation generation, meeting summaries, and knowledge repositories
  • Effective communication practices within large, distributed teams
  • Ethical use of AI in collaborative environments: ensuring fairness, transparency, and privacy

Module 8: Workshop and Practical Exercise with AI/LLM Integration

• • Interactive workshop on applying Agile and DevSecOps practices with AI Scenario:
    • Integrating AI/LLM workflows into the development lifecycle of a
      Satellite mission

Group activity

• • Designing an AI/LLM-assisted Agile sprint tailored to safety-critical system requirements
    • Discussion of outcomes, challenges, and mitigation strategies for AI integration

Biographies

Dr. Suzette Johnson works for Northrop Grumman Corporation near Baltimore, Maryland. As an NG Fellow for Lean Agile and Digital Integration she drives continuous improvement and modern ways of working to improve operational and program excellence to achieve mission outcomes. Her experience with Lean Agile began twenty years ago spanning across IT systems and software and systems engineering for cyber-physical systems. She leads transformation and improvement initiatives across the enterprise and government programs. She is serving as the NDIA Systems Engineering Division, Chair, to promote the widespread use of systems engineering in the Department of Defense acquisition process to achieve affordable and supportable solutions. She received a Doctor of Technology Management at the University of Maryland with a dissertation focused on investigating the Impact of leadership styles on software project outcomes in traditional and agile environments. As a respected thought leader, she has co-authored multiple papers and an award-winning book on Industrial DevOps, addressing how to integrate Lean, Agile, DevOps, and digital capabilities in cyber physical systems.

Description of Intended Audience and Recommended Prerequisites

Prerequisites: Open Mind; Basic Understanding of Agile and DevSecOps, basic usage of LLM is helpful

What can Attendees Expect to Learn

• • Understand the challenges, opportunities, and AI/LLM integration in applying Agile and DevSecOps practices to safety-critical cyber-physical systems at the system level, including governance for AI-assisted workflows.
  • Learn how to adapt Agile for large-scale system development. Become familiar with leveraging AI/LLMs to enhance efficiency, maintain rigorous safety standards, and improve backlog management, risk assessment, and compliance tracking.
  • Gain knowledge on integrating security, compliance, and AI into the development lifecycle, understanding how AI-driven tools can streamline threat modeling, automate compliance verification, and support decision-making in safety-critical environments.
  • Develop strategies for continuous integration, delivery, and deployment in highly regulated environments using CI/CD pipelines while balancing speed, precision, and safety.
  • Master systems thinking to manage complexity and interdependencies in cyber-physical systems development, leveraging AI for advanced modeling, simulation, and architecture analysis to design dependable and resilient systems.
  • Acquire practical experience in designing and implementing AI-assisted Agile sprints through hands-on workshops, applying scalable DevSecOps techniques to real-world industrial challenges, including ethical considerations in AI deployment.