Capstone Project — Full Stack + AI
SmartBoardAI
An AI-powered project planning application that turns plain-language ideas into structured, actionable project plans — built end to end by a four-person team for my DePaul University capstone.
Role
UX, Frontend, Team Lead
Context
DePaul Capstone
Timeline
Fall 2025
Stack
React, Java, Python, MySQL
The problem
Project planning tools require users to manually create, categorize, and prioritize every task from scratch. For students, freelancers, and small teams, this setup cost creates real friction before any actual work begins.
We wanted to build something that could take a plain-language idea and turn it into a structured, prioritized project plan in minutes — removing the setup barrier entirely and letting people focus on execution.
What we built
AI project generation
Users describe an idea in plain language. The AI middleware interprets the input and generates tasks, priorities, and milestones automatically through a conversational interface.
Interactive Kanban board
A drag-and-drop board with real-time status updates, color-coded priorities, and category tags. Built in React with live synchronization across team members.
Full-stack architecture
React and TypeScript frontend, Java Spring Boot backend, Python FastAPI AI middleware, and MySQL database — each layer designed for modularity and scalability.
Team collaboration
Real-time project synchronization, shared Kanban visibility, and activity tracking built for multi-user teams with secure JWT authentication.
My contribution
I contributed across the full stack — from UX strategy and interface design through to frontend implementation and team coordination. With a four-person team and a tight academic timeline, everyone wore multiple hats, and I took ownership of the product experience end to end.
On the UX side I defined the user flows, designed the conversational AI interface, and shaped how the Kanban board would present AI-generated output. On the engineering side I built the React frontend, integrating the AI middleware responses into the board UI in real time.
The UX documentation is being recreated in Figma as a formal case study — the product shipped, but the design artifacts are ongoing work.
What I learned
Shipping a full product in an academic sprint taught me a lot about prioritization under real constraints. Design documentation got deprioritized in favor of shipping — a tradeoff I understood at the time and am now correcting.
More importantly, SmartBoardAI confirmed something I already believed: the best product experiences come from people who understand both sides of the stack. Designing the AI interaction layer and then building it myself meant the final product matched the intent in ways that are hard to achieve across a handoff.
See it in action
View on GitHub →Frontend
React, TypeScript, Vite
Backend
Java Spring Boot, MySQL, JWT
AI Middleware
Python FastAPI, Together.AI SDK