Nova Campus Copilot
Nova Campus Copilot helps students stay on top of the college decisions that matter most when the inbox gets messy. Scholarship teams, financial-aid offices, housing portals, and admissions offices all send time-sensitive updates through different channels, and a single missed deadline can cost funding, housing, or enrollment momentum.
The app turns that chaos into a grounded action plan. It ingests a realistic student inbox, identifies the messages that actually matter, ranks the next actions by urgency and consequence, highlights the exact evidence behind each recommendation, and drafts polished follow-up replies the student can send immediately. Instead of making students read six long messages and guess what is critical, Nova Campus Copilot shows the top risks, the next deadline, the tasks that must happen first, and the reply draft needed to unblock each issue.
This project fits the Amazon Nova challenge because the product is built around a Nova-ready reasoning boundary. The live app uses a deterministic contract-compatible demo mode so the experience stays stable for judges, but the core inference path is already wired for Amazon Nova on AWS Bedrock through a dedicated adapter. That means the same UI, evidence model, task schema, and reply-draft flow can run against Amazon Nova without changing the rest of the product.
I built the app with Next.js, TypeScript, Zod, Vitest, and a Bedrock adapter layer for Amazon Nova. The product includes a scenario editor so a judge can change the wording or deadline inside a message and immediately rerun the analysis to see the priority stack reorder itself. That makes the demo feel live instead of scripted. The app is deployed publicly, the repository is public, and the submission includes a narrated walkthrough showing the product flow from inbox clutter to action-ready triage.
The main outcome is clarity under pressure. Nova Campus Copilot protects students from the kinds of administrative misses that quietly change life outcomes: losing a finalist interview slot, delaying aid, or letting a housing hold expire. It uses grounded evidence instead of vague advice, which makes the recommendations easier to trust and act on.
Built With
- amazon-nova
- aws-bedrock
- next.js
- typescript
- vercel
- vitest
- zod
Log in or sign up for Devpost to join the conversation.