Inspiration
The original inspiration for OpenFunds was to simplify and decentralize fundraising using blockchain technology. With the latest additions, we have extended this vision further by integrating deep learning, authentication, and containerized deployment. OpenFunds is now designed to be not only accessible and decentralized but also intelligent, secure, and production-ready.
We recognized the need for smarter systems that help users avoid campaign duplication, provide success predictions, and ensure that only authorized users can create campaigns. By merging artificial intelligence with decentralization, OpenFunds is evolving into a next-generation crowdfunding platform.
What it does
OpenFunds is a decentralized, AI-powered fundraising platform that enables users to: • Register and log in securely through a professional authentication system backed by SQLite • Create campaigns with detailed information such as title, description, target amount, wallet address, and owner name • Accept Bitcoin donations directly to their wallet (Testnet) • View and manage all campaigns in real-time • Receive predicted success likelihood and personalized feedback using machine learning • Discover similar past campaigns through a deep learning recommender • Access a full landing page styled like a modern company website before accessing the dashboard • Run entirely via Docker and Kubernetes with production-ready deployment
How we built it (Extended)
Frontend • Built using Streamlit with customized HTML and CSS • Landing page designed as a modern, company-level introduction to OpenFunds • Login and Signup system with form validation, session tracking, and database integration
Machine Learning • ml_predictor.py implements Logistic Regression and Random Forest classifiers • Models predict the success probability of campaigns based on key input features • Provides real-time feedback to campaign creators with improvement suggestions
Deep Learning • dl_similarity.py uses SentenceTransformers for semantic similarity • Automatically suggests similar campaigns in real-time based on title and description before submission • Fully local inference without the need for external APIs
Backend • SQLite database stores campaign and user login data • Campaigns are also written to CSV for compatibility with the DL model • Project is modular with separate directories for models, database logic, and Streamlit pages
DevOps • Dockerized the entire application for portability and reproducibility • Pushed the image to Docker Hub for easy sharing and cloud use • Deployed on Kubernetes using YAML-based configuration for deployment and service • Runs on Minikube or other local clusters for testing
New Challenges we tackled 1. Built DL recommender with no external APIs or keys using local models 2. Designed a custom login/signup workflow despite Streamlit’s lack of native authentication 3. Ensured that DL embeddings update immediately after each new campaign is created 4. Managed port forwarding and pod status effectively in Kubernetes 5. Handled Python dependency conflicts during Docker containerization
Accomplishments we’re proud of 1. Smart campaign similarity recommendations using deep learning 2. End-to-end AI pipeline using fully local ML and DL models 3. Secure authentication and login system with SQLite backend 4. Clean and modular multi-page application in Streamlit 5. A production-style landing page with full branding and UX clarity 6. Complete containerization and orchestration using Docker and Kubernetes
What we learned • How to use SentenceTransformers to implement fast and accurate similarity search locally • How to manage user authentication and page access control in Streamlit • How to integrate SQLite with both ML and DL data pipelines • How to containerize Streamlit apps and ML workflows using Docker • How to deploy applications on Kubernetes clusters from scratch
What’s next for OpenFunds 1. Integration of Metamask or other Web3 wallets for decentralized login 2. Support for Bitcoin mainnet transactions 3. Addition of Ethereum and ERC-20 token donation support 4. Hosting the app on decentralized storage networks such as IPFS 5. Admin panel for verifying campaigns and managing suspicious activity 6. Public campaign owner profiles with verified activity 7. Improved recommendation system using LLMs and advanced NLP 8. CI/CD pipeline using GitHub Actions for continuous deployment
OpenFunds has transformed from a simple decentralized crowdfunding tool into an intelligent, modular, and production-ready platform. It demonstrates how modern AI techniques, when combined with secure architecture and containerized deployment, can redefine the future of fundraising.
Log in or sign up for Devpost to join the conversation.