bug-triage-ai

This project is an AI that can read bug reports and go:

“okay this bug is serious, kinda urgent, and send it to team backend pls 😤”

basically it helps devs by tagging bugs automatically.

---

what’s the point?

bugs are annoying. figuring out where it is also annoying.

so Imma use some machine learning tech:

• guess the type
• guess how severe it is
• guess the priority
• guess which team should fix it

---

tools used

• Python
• pandas + scikit-learn for data & ML
• sentence-transformers (BERT, honestly I'm still clueless the full function of it, still learning)
• MultiOutputClassifier with LogisticRegression
• joblib to save
• pytest making sure nothing explodes
• streamlit to show off (optional 👀)

---

how it supposed to work

1. generate_bugs.py → makes fake bug reports and saves bugs.csv
2. bug_classifier.py → processes data, evaluates it
3. bug_model.py → use ML logic (still messy)
4. utils.py → text cleaning, etc.
5. app.py → streamlit app for live prediction demo
6. tests/ → testing the function works or not

it predicts these:

• label → what kind of bug? (UI, Backend, etc)
• severity → how bad is it?
• priority → how fast do we fix this?
• team → who’s in charge?

---

getting started

# clone it
git clone https://github.com/Faiz2468/bug-triage-ai.git
cd bug-triage-ai

# make the environment
python -m venv venv
venv\Scripts\activate   

# install all ingredients
pip install -r requirements.txt

# create some bugs
python data/generate_bugs.py

# train the brain
python bug_classifier.py

# run the UI
streamlit run app.py

## MIT licence
Do what u want. just don’t sell it