Project Story: Fundara - AI Compute Crowdfunding

About the Project:

Fundara is an AI crowdfunding platform designed to address the critical need for funding in the rapidly expanding field of AI research. Similar to Kickstarter, but specifically tailored for AI researchers, Fundara allows researchers to request funding for compute resources (GPUs, cloud credits, etc.) needed for their projects. Investors can then browse projects, invest using traditional methods or Solana, and follow progress through integrated updates. We envisioned integrating with Hugging Face and Weights & Biases to showcase model performance and research outputs directly within the platform.

Inspiration:

The inspiration stemmed from the booming AI landscape and the significant financial barrier to entry for many researchers. Securing sufficient compute power is often a major hurdle, limiting innovation and accessibility. Fundara aims to democratize access to resources, fostering a more inclusive and vibrant AI research community. We believe that by providing a dedicated crowdfunding platform, we can encourage smaller-scale, potentially more creative and innovative AI projects that might otherwise lack the necessary funding.

Development Process:

We utilized a tech stack comprising Next.js for the frontend, MongoDB for data storage, and Clerk for user authentication. Stripe was integrated for traditional secure payment processing, and we also integrated Solana for other less traditional payment methods. Our development followed a phased approach:

1. Foundation: We started with a basic Next.js project structure.
2. Authentication: Clerk was integrated to handle user accounts and authentication securely.
3. Backend: MongoDB was set up and connected to the frontend, providing a robust backend for project and user data.
4. Frontend Development: We built various pages, including a homepage showcasing projects, individual project pages with details and updates, a project creation page, a user dashboard for managing projects and donations, and a secure login/signup system.
5. Payment Integration: Stripe's API was integrated on both the frontend and backend to enable secure donations. We also integrated Solana's API so users can pay straight from their crypto wallets for donations.
6. AI Integration (Partial): We implemented a basic integration with Weights & Biases and Hugging Face, showcasing initial progress towards our vision of fully integrated model performance tracking.

Lessons Learned:

This hackathon provided valuable learning experiences:

• Planning is Key: While we had a general plan, a more detailed roadmap would have improved efficiency and focus. Exploring multiple ideas simultaneously proved less effective than sticking to a core vision.
• Time Management: Coding projects often take longer than anticipated. We underestimated the time required for debugging and implementation, highlighting the need for more realistic time budgeting in future projects.
• Technical Skills: We gained significant experience with Stripe, MongoDB, and integrating AI tools. This hands-on experience significantly expanded our skillset.

Challenges Faced:

The primary challenges revolved around debugging unfamiliar technologies. Our limited experience with Stripe and Solana among other technologies resulted in some lengthy debugging sessions. Time constraints and fatigue also played a role, impacting our overall progress.

In conclusion, Fundara represents a significant step towards creating a more accessible and inclusive environment for AI research. While we faced challenges, the lessons learned and the progress made are invaluable.

Built With

• clerk
• huggingfaceapi
• javascript
• mongodb
• next
• stripe
• typescript
• weightsandbiasesapi