Hi, I'm Nina!

Problem: Machine learning models often degrade silently as social media trends and user behaviors shift over time.

Solution: I built an automated MLOps pipeline that monitors statistical data drift and tracks model performance using MLflow and Evidently AI.

Problem: Overcrowded gyms are annoying

Solution: Send me notifications when gym attendance drops below a set number during specific hours.

Problem: It takes me too long to find a new book to read and also I wanted to learn more about SQLAlchemy and Pydantic

Solution: Built Python + SQLAlchemy backend as a sandbox for recommendation ideas, NLP practice, and database design, currently using LLM-based recommendations...

Problem: Detecting emotions in text messages requires context and careful evaluation across approaches.

Solution: Built and evaluated Naive Bayes, Logistic Regression, LSTM (Keras), and BERT with preprocessing/analysis in pandas and visualisation in Matplotlib.