Software Project # 1 [BuildIT] 🚨 The Problem As we're building in this hackathon we noticed a major gap, hardware building and prototyping is still stuck in the prompt era. Today, if you want to build something physical, you have to: Translate messy ideas into precise instructions Manually reason about wiring, constraints, compatibility, and sequencing Jump between datasheets, YouTube tutorials, and trial-and-error Prompt AI repeatedly just to keep context alive This is the core gap: AI understands language. Hardware requires structure, memory, and physical reasoning. No existing tool bridges that gap.
What it does
Hardware Builder is a web app that helps you prototype electronics projects end-to-end: Parts Library (40+ parts listed as a core feature) Two workflows Parts → Build: “Here’s what I have, what can I build, and how?” Idea → Shopping: “Here’s what I want, what do I need to buy?” AI build planning via OpenRouter Component generation and image recognition via Gemini Interactive 3D assembly viewer for guiding how a build comes together Project persistence with MongoDB (cloud sync) Offline-first fallback: if the backend is unavailable, the app falls back to localStorage Downloadable Arduino code + test checklists with progress tracking
How we built it
Frontend React single-page app (main UI lives in App.js) Talks to: OpenRouter API for build-planning responses Gemini API for image-based component recognition Uses localStorage fallback for offline/local mode when the backend is down
Backend Node.js + Express API server (server.js) Responsibilities (as described in README): Save/retrieve projects in MongoDB Atlas Store parts inventory Handle image uploads to DigitalOcean Spaces (optional) Data + Storage MongoDB Atlas for persistence (projects + inventory) DigitalOcean Spaces for optional cloud image storage (S3-compatible object storage)
Challenges we ran ihnto
Designing something that still works when a laptop can’t stay online: the app needs to degrade gracefully.
“AI planning” is only useful if it’s grounded in real constraints (parts on hand, wiring realities, testability), so the UX needed structure, not just chat.
Accomplishments that we're proud of
A complete pipeline that blends: parts inventory AI planning vision-based component recognition persistence + offline fallback build artifacts like Arduino code + test checklists
What we learned
How to integrate MongoDB, GEMINI, Openrouter and DigitalOcean
What's next?
Make the 3D viewer more “assembly-true”: snapping, collision checks, and breadboard-aware constraints. Add “hardware sanity checks”: current draw estimation, pin conflicts, and known wiring pitfalls. Exportables: BOM, wiring diagram, and step-by-step assembly cards.
https://github.com/sshakerinezhad/BuildIT (hosted on DigitalOcean) UTRA Challenge # 2
Our BuildIT tool helped make our bot that's going allllll the way to the olympics!!!!
Log in or sign up for Devpost to join the conversation.