elevator

it’s about solving problems with AIReal problems.From prompt to prototypeAI that actually ships it’s about teams learningWhere builders meet brains Hack today.

Comment

1. What inspired us We were inspired by problem you saw because why it matters to you/people. We wanted to see if AI could solve it in a new way

2. What we learned We learned how to tech skill how to work with API/model/data and that biggest takeaway about AI/product/teams

3. How we built it We built project name using tech stack e.g. Python, OpenAI API, React, FastAPI. The flow is: data in → AI model does X → output to user We trained/fine-tuned/connected it in X hours

4. Challenges we faced Our biggest challenges were technical challenge] and team/time challenge. We solved it by what you did If we had more time we’d add next feature

Example for an AI Hackathon

What inspired us We were inspired by students in Tehran struggling to study for exams because good tutors are expensive. We wanted to see if AI could act like a 24/7 personal tutor.

What we learned We learned how to use LLMs with RAG to keep answers accurate, how to deploy a model on a budget, and that prompt engineering is 50% of the work.

How we built it We built StudyBuddy AI using React, FastAPI, and GPT-4 with vector search. Students upload notes → our AI summarizes them + generates quiz questions → gives instant feedback. We built the MVP in 36 hours.

Challenges we faced Our biggest challenge was hallucinations giving wrong answers. We fixed it by grounding responses in the uploaded notes. Another challenge was API rate limits, so we added caching. Next we’d add voice and support for different languages

  1. *Project name +Hackathon
  2. *Tech you used*pav22centfutureAI
  3. *1 big challenge*very inexplicable
Share this project:

Updates

posted an update —

Elevator" Project Updates*

Post Title: [v1.2.0] Elevator Update: AI Tutor + Mobile Release

Body:

Update #3: Elevator is getting smarter

What’s new:

  • New Feature: Added AI Quiz Generator. Upload notes → get instant practice questions
  • App Store: Elevator is now live on iOS! Download here: [App Store Link]
  • Improvement: 40% faster response time by switching to GPT-4o-mini
  • Bug Fix: Fixed file upload crash on Android

Screenshot: [attach image] Code Snippet:

def generate_quiz(notes):
 return llm.create_quiz(context=notes)

What we learned this week: RAG works way better when we chunk by section, not page.

Next up: Voice mode + language support 

### **3 Example Update Posts for "Elevator"**

#### **1. Launch Update**

 Elevator v1.0 is LIVE!

We built Elevator to make studying less painful. Upload any PDF and our AI turns it into summaries + quizzes.

New in this release:
- PDF Upload + AI Summary
- Web App: [elevator.app]
- GitHub: [github.com/yourteam/elevator]

Built with: React, FastAPI, GPT-4o
Shoutout to the team for 36 hours straight 

Try it and tell us what you think!


#### **2. Feature Update**

 Update #2: Elevator now has Memory

Your AI tutor now remembers past chats so it can reference what you learned last week.

New:
- Chat History + Memory
- Dark Mode
- Export notes to PDF

Challenge we hit: Memory was making responses slow. Fixed with vector caching.

Testing the beta? Drop feedback here.
**3. App Store Release**

Elevator is on the App Store!

After 2 months of building, you can now study on the go.

Download: [App Store Link] | [Play Store Link]

What's inside v1.1:
- Mobile app for iOS + Android
- Voice Q&A with Whisper
- Offline summaries

Code for the mobile app: [GitHub Link]

We’ll be reading every comment and review. Thank you for the support!

### **Tips to get engagement:**
1. **Show, don’t just tell**: Add 1 screenshot or 10s screen recording
2. **Be specific**: "40% faster" > "made it faster"
3. **Ask a question at the end**: "What feature should we add next?"
4. **Post regularly**: v0.1 → v0.2 → Launch → Post-launch

Log in or sign up for Devpost to join the conversation.