What I Built
NexusFlow is a production-grade, real-time Agile collaboration platform inspired by tools like Trello and Jira. I originally started building it as a full-stack portfolio piece to challenge myself with WebSockets (SignalR) and optimistic UI patterns.
What makes NexusFlow special is its deep AI integration: it can generate entire Kanban boards from a simple prompt, break down vague ideas into actionable subtasks, and provides an AI writing assistant directly inside the rich-text editor. It means a lot to me because it successfully brings together a complex stack (.NET 9, React 19, Entity Framework Core, PostgreSQL) into one seamless, real-time experience that actually feels good to use.
Demo
🚀 NexusFlow
A real-time, full-stack Agile collaboration platform built with .NET 9 and React 19 - with deep AI integration powered by OpenRouter.
About • Demo • Features • Architecture • Tech Stack • Getting Started • Diagrams • Project Structure • Issues & Feedback
About
NexusFlow is a production-grade project management tool inspired by Trello and Jira, built from the ground up as a full-stack portfolio project. It features real-time collaboration via SignalR WebSockets, a rich Kanban board with drag-and-drop, multiple view modes, and a deep AI layer powered by the OpenRouter API.
Demo
🌐 Live Demo
Try it out: https://nexusflow123.vercel.app
⚠️ The live demo runs on free-tier infrastructure (Vercel + Render + Supabase). The backend may be slow to respond on the first request (cold start ~30s) and could occasionally be unavailable. Performance will improve after the first request wakes the server up.
Real-Time Collaboration (SignalR)
Signal.R.mp4
AI
…You can check here the demo!
The Comeback Story
I started NexusFlow in February 2026 with an ambitious vision, but like many side projects, I struggled to push it over the final hurdle of deployment and production readiness. It sat locally on my machine as a "mostly working" app, but configuring cloud hosting, fixing complex CORS and SPA routing issues across a monorepo, and containerizing the database seemed daunting and time-consuming.
For the Finish-Up-A-Thon, I finally dusted it off and decided to make it live. Here is what I fixed and added to finish it up:
Cloud Deployment & CI/CD: Configured a full deployment setup using Vercel (Frontend), Render (.NET Backend), and Supabase (PostgreSQL).
Networking Fixes: Resolved tricky cross-origin (CORS) and WebSockets (SignalR) routing issues for the production environment so the real-time collaboration actually works securely online.
Developer Experience: Added comprehensive setup scripts and documentation to allow anyone to spin the project up locally via Docker with a single command.
UI Stabilization: Fixed various UI bugs and stabilized the drag-and-drop feature to prevent ghost-card artifacts when multiple users are editing the board simultaneously.
My Experience with GitHub Copilot
GitHub Copilot was an absolute game-changer in helping me finish this project, especially when tackling the final deployment hurdles. When I was stuck on the complex SPA routing configurations for Vercel (vercel.json) and configuring my CORS headers in .NET 9, Copilot was able to contextually analyze my project structure and suggest the exact configurations needed to make the frontend communicate securely with my backend and SignalR hub.
Copilot also accelerated writing the boilerplate for my Entity Framework Core deployment scripts and SQL migrations, allowing me to focus on the high-level architecture rather than database syntax. Finally, whenever I hit a roadblock with React 19's new features or debugging drag-and-drop edge cases, Copilot Chat provided instant, contextual debugging steps that saved me hours of frustration.
Top comments (0)