Inspiration
At WashU, we realized that students are surrounded by opportunities—research labs, clubs, career events, and programs—but many don’t know where to start. The problem isn’t a lack of resources, but a lack of clarity and direction.
From our own experiences, especially as early-stage students, we often asked:
Where should I begin? What should I do next? Which opportunities are actually right for me?
While tools like ChatGPT can provide general advice, they don’t understand the structure of a specific university or guide students toward real, actionable steps.
This inspired us to build a system that doesn’t just answer questions—but helps students make decisions and take action.
What it does
WashU Engineering Plan Builder is an AI-powered platform that transforms vague student goals into personalized, actionable plans.
Instead of overwhelming users with information, it provides:
- Top recommended resources tailored to the student’s goal
- Proposed next steps (clear and actionable)
- Links of the resources that fits you
For example, a student asking:
“I am seeking for an internship for embedded systems, where should I start?”
receives not just advice, but a prioritized plan grounded in real campus resources.
How we built it
- Backend:
- Hosted on Render
- Built with Python (FastAPI/Flask)
- Integrates OpenAI API for reasoning and plan generation
Uses structured datasets (scraped + curated) including: research opportunities, faculty directories, campus resources
Frontend:
Built using Lovable and deployed via Vercel
Modern, responsive UI with: Goal input form, dynamic result rendering, and loading and error states.
Core logic:
Resource matching layer (rule-based filtering)
AI layer that converts: Goal + User Context + Resources → Action Plan
This hybrid approach ensures outputs are both relevant (data-driven) and intelligent (AI-generated).
Challenges we ran into
- Data fragmentation: WashU resources are spread across many pages, making scraping and structuring difficult.
- Backend errors (500 issues): Handling missing environment variables and unexpected API failures required robust validation and error handling
- Frontend–backend integration: Debugging API calls across environments (preview vs deployed) was tricky
- Avoiding “just another ChatGPT”: We had to rethink the product to focus on decision-making and action, not just information
Accomplishments that we're proud of
- Built a working end-to-end system from scratch
- Designed a product that is meaningfully different from ChatGPT
- Created a real-world use case with immediate student impact
- Successfully integrated: Data pipeline, AI reasoning, and frontend interaction
Most importantly, we turned an abstract idea into a functional, demo-ready product.
What we learned
- Building with AI is not just about calling a model—it’s about designing useful outputs
- Data quality and structure matter more than model complexity
- Debugging real systems (APIs, deployment, environment variables) is a key engineering skill
- Product thinking is critical: The question isn’t “Can we build this?”, it’s “does this actually help users?”
What's next for Washu Engineering Plan Builde
We see this as just the beginning. Next steps include:
- Expanding the dataset to include: Resources for other colleges in Washu, Clubs, fellowships, internships, and alumni pathways
- Improving personalization using: Major, year, and user history
- Adding features like: “Students like you” recommendations, opportunity tracking and reminders
- Scaling beyond WashU: Adapting the system to other universities
Our long-term vision is to build an AI-powered navigation system for student success, helping anyone go from: “I don’t know what to do” → “Here’s exactly what I’ll do next”
Log in or sign up for Devpost to join the conversation.