Swarag Bk

I've built machine learning and deep learning models at Fututa Labs Technologies for classification and sentiment analysis, processing more than 50,000 text and image data points with Python, NLP, and data augmentation.

I developed XGBoost, Scikit-learn, TensorFlow, and PyTorch solutions that achieved 88–92% accuracy. My work across preprocessing, feature engineering, and model evaluation reduced errors by 20–30%.

For an Air Pressure System fault-detection project, I streamlined an XGBoost model on 60,000+ records to reach 0.92 ROC-AUC and improve performance by 18% over a Logistic Regression baseline. I reduced dimensionality by 35%, improved training speed by 40%, and used SMOTE and hyperparameter tuning to raise recall by 20%.

Before moving into data science, I supported production applications at Infosys, maintaining 99.9% uptime and resolving 100+ incidents using SQL, ServiceNow, Jira, and root-cause analysis. I bring this production mindset to scalable, reliable machine learning solutions.