Overview

We created a leaderboard that could be used throughout an ML hackathon so participants could accurately see their current placement. From there, we adapted it to work with generalized hackathons — specifically sponsored tracks.

Difficulty — Beginner

Project members were not expected to have ML experience, but needed familiarity with coding.

Milestones

Month one — planning and initial setup. Define project goals and scope, including requirements for both generalized and sponsored hackathons. Finalize the technical stack for frontend, backend, and MLflow integration, and set up the development environment and version control. Create wireframes and mockups for the leaderboard UI and begin the backend architecture, including database schema design. Hold a kickoff meeting, finalize the execution strategy, and start initial development.

Month two — development and integration. Develop core frontend components such as leaderboard display, user registration, and login, while building backend APIs for data retrieval and user management. Integrate MLflow for tracking and logging model performance metrics during the hackathon. Implement features specific to sponsored hackathons, such as track-specific leaderboards and custom scoring. Conduct internal testing focused on data flow between frontend, backend, and MLflow, and begin documenting the codebase.

Month three — testing, refinement, deployment. Conduct end-to-end testing with sample data simulating hackathon scenarios, including edge cases for sponsored tracks. Gather stakeholder feedback and iterate. Optimize the backend for scalability so it handles large data volumes during the hackathon. Prepare deployment: hosting, database migration, and CI/CD pipelines.

Month four — deployment and post-deployment. Deploy to a test environment and conduct a soft launch with selected participants. Officially launch for a live hackathon, supporting participants and gathering feedback during the event. Afterward, analyze performance, document lessons learned, clean up the codebase, and prepare for handoff.

Team

Four to eight members, split into subteams by skill set, each responsible for managing their allocated task. Meetings twice weekly, dropping to once weekly when progress was smooth.