Sprint Pulse

Sprint Pulse: Empowering Teams, Preserving Well-being! Sprint Pulse predicts employee burnout levels to ensure mental well-being and productivity.

Comment

TLDR; Watch the short 5 min video for the Intro, Story, and Live Demo.

Inspiration

Sprint Pulse was born out of our own experiences working in agile teams, where sprint planning often overlooks the impact of unexpected ad-hoc tasks. Receiving a flood of last-minute assignments can be mentally exhausting, leading to burnout. Despite this, we found no tools that could identify and address this issue proactively. This app is our effort to fill that gap and support teams by emphasizing mental well-being while maintaining productivity.

What it does

Sprint Pulse is a real-time app that predicts employee burnout levels by analyzing:

  • Current Sprint Metrics: Planned vs. ad-hoc tasks and their assignment timings.
  • Sprint History: Patterns of task allocations and workload consistency.
  • Burnout Likelihood Prediction: Categorizing employees as "No Burnout," "Hardly Likely," "Likely," or "Definitely."

The app seamlessly integrates with Jira to fetch data, providing actionable insights to help teams stay balanced and productive.

How we built it

  • Frontend: Atlassian Forge UI Kit with Node.js v20 and React.
  • Data Generation: Synthetic data creation, including features such as:
    • Day of ad-hoc task assignment.
    • Total planned story points.
    • Burnout patterns based on sprint history.
  • Backend:
    • AI Models: Random Forest for burnout classification, fined tuned using Grid Search Algorithm.
    • Flask APIs for real-time communication between the frontend and backend.
  • Deployment:
    • App: Hosted on Atlassian's Forge Cloud Serverless Deployment Platform.
    • API: Hosted on PythonAnywhere.
  • Integration: Seamlessly integrated with Jira using Atlassian Forge CLI and REST APIs.

Challenges we ran into

  • Dataset Unavailability: With no open-source datasets for burnout prediction, we had to generate synthetic data, incorporating key factors like the timing of ad-hoc task assignments and total planned story points.
  • Real-Time Data Processing: Ensuring smooth real-time functionality without persistent storage.

Accomplishments that we're proud of

Building an AI-driven solution that addresses a critical gap in team management tools. Seamlessly integrating with Jira while maintaining a lightweight, serverless architecture.

What we learned

  • The value of synthetic data in model development when real-world data is unavailable.
  • To use Jira APIs and the Forge platform. Also, how easy and seamless it can be to create and integrate an app using it to meet specific needs.

What's next for Sprint Pulse

  • Collaborations for Real Data: Working with Atlassian to enhance model training with real-world datasets.
  • Advanced Analytics: Introducing features for workload redistribution and team-level insights.
  • Wider Integration: Expanding support to additional project management platforms.
  • Improved Scalability: Optimizing the app for larger teams and enterprises.
Share this project:

Updates