Inspiration
Individually reviewing a software project's pull requests, issues and active alerts can be hectic.
Bitfull of Work is an AI web app built to ease the process by using Machine Learning to detect the most critical pull-requests and project issues.
What it does
Bitfull is an AI powered web app that uses Machine Learning to determine the most relevant Bitbucket project issues and pull requests that require the developer's urgent attention.
How we built it
The app uses a pretrained Machine Learning model that is fine-tuned to identify genuine in-progress Bitbucket project issues, open pull requests, and active alerts, while filtering out profanity, abuse and irrelevant pull-requests and project issues.
Bitfull then plots a graph to notify the developer about the most critical project issues, pull requests and alerts that are worth the developer's time.
The app uses Pandas and Scikit-Learn to build a Machine Learning Profanity Detection Model that also filters out un-important pull-requests and issues.
For example 'README.md' documentation updates are also ignored by the model, because they aren't critical updates.
LIVE DEMO: https://bitfull.myeulome.com
SOURCE CODE: https://github.com/VakinduPhilliam/bitfull
BITBUCKET PROJECT LINK: https://bitbucket.org/vakindu/bitfull/src/main/
Challenges we ran into
The Bitbucket API is quite difficult to learn for beginners. So is the adoption hurdle for bitbucket API and Forge Authentication.
We also had difficulty choosing the right Machine Learning Model to use for the web app.
Accomplishments that we're proud of
The AI powered project review app can reduce developer's pull-request review time by atleast 40%.
The app's statistical graph features also help developers better manage time spent fixing project issues.
What we learned
We gained a lot of experience working with Jira, bitbucket, and Forge.
Jira is a great platform for building large world-class software projects.
What's next for Bitfull
Working on new versions for iOS and Android.
Also working on in app purchases.
Log in or sign up for Devpost to join the conversation.