SWE Cloud Agent Workshop

Build a complete cloud-based Software Engineering agent system: a Codex GPT-like frontend showing tasks/status/logs + customizable agent backend with deployment options.

Architecture Overview

Frontend: Build your own Codex GPT replica displaying:

  • Real-time task execution status
  • Agent activity logs and monitoring
  • Code generation results
  • Repository management interface

Backend: Leverage xpander.ai’s Backend-as-a-Service for agents:

  • Framework-agnostic infrastructure
  • Custom agent loops with any LLM model
  • Local function execution
  • Serverless → self-hosted deployment path

xpander.ai Backend-as-a-Service Benefits

ComponentTraditional Setupxpander.ai BaaS
InfrastructureManual server setupManaged, auto-scaling
State ManagementCustom implementationBuilt-in distributed state
Tool IntegrationManual API wrappersMCP-compatible library
Real-time EventsWebSocket implementationEvent streaming included
Multi-user SupportComplex auth & isolationBuilt-in multi-tenancy
DeploymentDevOps overheadOne-command deployment

What You’ll do during this workshop

  • Use xpander.ai built-in frontend: Chat interface to to talk with you agent
  • Task Management Frontend: React dashboard for SWE agent operations
  • Agent Backend: Customizable loops with OpenAI Codex integration
  • Cloud Deployment: Production-ready system with monitoring
  • Competition Submission: Package for apps-by-agent competitions

xpander.ai Backend-as-a-Service

Framework-agnostic infrastructure for autonomous agents:

FeatureDescription
🛠️ Framework FlexibilityOpenAI SDK, Agno, CrewAI, LangChain, or native LLM APIs
🧰 Tool IntegrationMCP-compatible tools library and pre-built integrations
🚀 Scalable HostingManaged infrastructure with auto-scaling
💾 State ManagementDistributed state with multi-user support
Real-time EventsAgent2Agent, Slackbots, WebUI, Webhooks
🛡️ API GuardrailsAgent-Graph-System for tool dependencies

Workshop Modules

Prerequisites

  • Development: Python 3.12, Node.js v22, React experience
  • Backend: xpander.ai account, OpenAI API access
  • Integration: GitHub account with repository permissions

Key Learning Outcomes

  • Agent Architecture: Design patterns for production SWE agents
  • Frontend Development: Real-time dashboards for agent monitoring
  • Backend Customization: Custom agent loops and LLM model switching
  • Deployment Strategies: Serverless → self-hosted migration paths
  • Competition Readiness: Professional submission packaging

Ready to build a production SWE agent system? Start with Module 1!