AI-Driven Zero-Code SW Engineering

OVERVIEW

AI-Driven Zero-Code SW Engineeringis an industry-led program for undergraduate Computer Science and Engineering students who want to build, version, test and deploy applications through natural-language AI agents, Git/GitHub and CI/CD without relying on routine manual coding.

The course follows a no-code-first, agentic workflow. Learners describe product requirements in natural language, while AI agents assist with planning, application scaffolding, feature development, debugging, refactoring, testing and documentation. Students remain responsible for reviewing plans, inspecting changes, validating outputs and approving actions.

Teams use Google Antigravity, OpenAI Codex and Claude Code within an industry-style delivery process. GitHub becomes the project source of truth, and every team completes a deployed product supported by repository history, automated quality checks, CI/CD evidence, release documentation and controlled rollback.

AUDIENCE PROFILE

• Undergraduate Computer Science and Engineering students
• Students in Information Technology, Software Engineering or allied computing disciplines
• Learners preparing for software engineering, web technology, cloud, DevOps, testing or project-lab work
• Students interested in natural-language application development and AI-agent workflows
• Aspiring AI application builders, junior product engineers, QA automation associates and DevOps associates
• Project teams seeking practical experience with repositories, pull requests, CI/CD and deployment
• Learners who want portfolio evidence for internships and entry-level technology roles

PRE-REQUISITES

• Undergraduate standing in Computer Science and Engineering or an allied discipline
• Basic awareness of web applications and the software development life cycle is helpful
• Programming proficiency is not mandatory
• Laptop, browser access and institutional access to the prescribed AI and GitHub tools
• No prior experience with agentic coding, the Git command line or CI/CD is required
• Willingness to inspect AI-generated changes, protect data and follow approval rules

COURSE TYPE

Industry-Led / Instructor-Guided Training with Applied Labs, Team Sprints, Repository Collaboration, Deployment Practice and Capstone Demonstration

DURATION

10 Sessions / 20 Hours

CURRICULUM FIT & CAREER READINESS

• Complements Software Engineering, Web Technology, Cloud Computing, DevOps, Testing and project-lab courses
• Recreates industry practice through product backlogs, sprint reviews, issue-to-pull-request flow, CI/CD and mentor feedback
• Replaces unverified copy-paste development with specification-driven, reviewable and reversible delivery
• Builds readiness for AI application builder, junior product engineer, QA automation and DevOps associate roles
• Produces portfolio evidence suitable for internships, technical interviews and project demonstrations

COURSE DESCRIPTION

The program begins with a product brief and converts it into specifications, user stories, acceptance criteria, screen flows, a basic data model and an agent-ready sprint backlog. Learners then use different AI coding agents for clearly defined roles while maintaining human ownership of all decisions and outputs.

Google Antigravity is used for workspace-based planning, execution and browser or terminal verification. OpenAI Codex supports repository onboarding, issue-to-task workflows, multi-file changes, tests and diff review. Claude Code supports project-context work, permissions, plan-first execution, debugging, refactoring, testing and documentation.

Git and GitHub provide the collaboration backbone through clone, pull, branch, commit, push, pull request, review, merge, conflict resolution, revert and release-tag workflows. Teams add automated testing, linting, dependency and secret checks, approval gates, GitHub Actions, preview deployment, monitoring and rollback. The final result is a reviewable, reproducible and maintainable team product rather than an unverified AI-generated prototype.

COURSE CURRICULUM

Module 1 - Agentic No-Code Engineering
• Industry software delivery life cycle, natural-language specification, AI-agent roles, prompting, validation and safe execution
• Human ownership, privacy boundaries, approval rules and an agent-use charter
• Output: Product brief and agent-use charter

Module 2 - Product and UX Blueprint
• User problem, business value, user stories, acceptance criteria, screen flow and data model
• Product requirements document, sprint planning and agent-ready backlog design
• Output: PRD, wireflow and sprint backlog

Module 3 - Build with Google Antigravity
• Workspace setup, agent planning, task execution and browser or terminal verification
• Application scaffolding, first working flow and artifact-based validation
• Output: Working MVP v0.1 and artifact log

Module 4 - Build with OpenAI Codex
• Repository onboarding, project instructions, issue-to-task flow and multi-file changes
• Test generation, execution evidence, diff review and feature-branch discipline
• Output: Codex feature branch and test evidence

Module 5 - Build with Claude Code
• Project context, permissions, plan-first workflow, debugging and refactoring
• Testing, documentation and structured bug-fix reporting
• Output: Claude feature and bug-fix report

Module 6 - Git and GitHub Collaboration
• Clone and pull, branches, small commits, push, pull requests, review and merge
• Conflict resolution, revert, repository traceability and release tags
• Output: Team repository and merged pull request

Module 7 - Quality, Testing and Secure Delivery
• AI-generated tests, linting, dependency checks and validation evidence
• Secrets, environment variables, privacy, permissions, approval gates and secure handover
• Output: Test suite and quality/security checklist

Module 8 - CI/CD and Preview Deployment
• GitHub Actions triggers, build and test jobs, artifacts and status evidence
• Branch protection, preview environments, deployment and pipeline troubleshooting
• Output: Green pipeline and deployed preview URL

Module 9 - Industry Release Sprint
• Process a change request, pull the latest version and implement approved changes with AI agents
• Diagnose CI failures, deploy, monitor, document release evidence and perform controlled rollback
• Output: Deployed v0.3 and release/rollback log

Module 10 - Capstone Demo and Handover
• Present product value, architecture, repository history, prompts, tests and pipeline evidence
• Explain limitations, responsible-AI controls, maintenance requirements and future roadmap
• Output: Deployed MVP, repository and seven-minute demonstration

TOOLS & PLATFORMS

• Google Antigravity for workspace planning, execution and artifact-based verification
• OpenAI Codex for repository onboarding, feature development, tests and diff review
• Claude Code for project-context work, debugging, refactoring, testing and documentation
• Git and GitHub for source control, issues, branches, commits, pull requests, reviews and releases
• GitHub Actions for build, test, lint, validation and deployment workflows
• Browser and terminal verification for application behavior and agent output review
• Preview or production deployment environment with approval and rollback controls
• README files, project instructions, prompt logs, test evidence and handover documentation

LEARNING OUTCOMES

Upon completion, learners will be able to:
• Build and deploy a functional application from natural-language requirements without routine manual coding
• Select and use Antigravity, Codex and Claude Code for planning, generation, debugging, testing and documentation
• Convert a product brief into specifications, user stories, acceptance criteria and an agent-ready backlog
• Maintain a shared repository through pull, branch, commit, push, pull request, review, merge, release and rollback
• Create CI workflows that automatically build, test, lint and validate approved changes
• Deliver updates to a preview or production environment and diagnose failed pipelines
• Apply secure and responsible AI engineering through code review, privacy, secret management and traceability
• Present evidence of human ownership, validation, testing, release control and technical handover

CAPSTONE PROJECT

Teams complete an industry-style capstone involving:
• A defined user problem, product brief, PRD, user stories and acceptance criteria
• A functioning application built and improved through approved AI-agent workflows
• A collaborative GitHub repository with issues, branches, commits, pull requests, reviews and releases
• Automated tests, linting, dependency checks and security or quality controls
• A working GitHub Actions CI/CD pipeline with preview or production deployment evidence
• A release and rollback log showing controlled change management
• A README, prompt/specification history and technical handover package
• A seven-minute demonstration covering value, architecture, evidence, limits and roadmap

PORTFOLIO DELIVERABLES

• Product Brief, PRD, User Stories and Acceptance Criteria
• Agent Prompt/Specification Log and Project Instruction Files
• Antigravity-Built MVP with Artifact-Based Verification
• Codex and Claude Code Feature Branches with Reviewed Diffs
• GitHub Repository with Issues, Commits, Pull Requests and Releases
• AI-Generated Test Suite and Security/Quality Checklist
• GitHub Actions CI/CD Workflow with Status Evidence
• Deployed Application with Release and Rollback Log
• Industry Capstone Package, README and Final Demonstration

ASSESSMENT APPROACH

• Guided labs, tool onboarding and completion of sprint tasks
• Quality of product specifications, prompts, reviewed changes and validation evidence
• Git/GitHub discipline through small commits, branches, pull requests, reviews and traceability
• Reliability of tests, CI/CD pipeline, deployment, security controls and rollback
• Industry mentor review, final product demonstration and technical handover

PROFESSIONAL HABITS DEVELOPED

• Specify before building and use acceptance criteria to guide AI agents
• Pull before work, isolate changes in branches and keep commits small and reversible
• Review plans, diffs, tests and commands before approving agent actions
• Never expose passwords, tokens, personal data or production secrets in prompts or repositories
• Treat CI failures as evidence to diagnose rather than checks to bypass
• Document decisions so another team can reproduce, maintain and extend the project

COMPLETION FOCUS & CAREER PATHWAYS

Students leave with a deployed team MVP maintained through Git/GitHub and an automated CI/CD pipeline. The final portfolio includes the PRD, prompt logs, repository history, pull requests, tests, pipeline runs, releases, rollback evidence and handover notes.

The course prepares learners for AI-native software delivery, internships and entry-level pathways such as AI Application Builder, Junior Product Engineer, QA Automation Associate and DevOps Associate. It also strengthens readiness for software engineering, web application, cloud, testing and project-lab work.