Tansiv Jubayer

I reduced energy forecasting error by 15% RMSE for my IEEE COMPAS 2025 bachelor’s thesis, using feature engineering and optimized ETL pipelines for hybrid LSTM–XGBoost models.

My research includes four peer-reviewed IEEE and Springer publications spanning sustainable energy forecasting, sugarcane disease detection, medicinal plant classification, and earthquake magnitude prediction. I’ve evaluated machine learning models with metrics including RMSE and MAE, improved classification precision by 10–12% using a Swin Transformer, and achieved 92% accuracy by fine-tuning ResNet50.

I also build stakeholder-ready Power BI sales and HR dashboards using DAX, Power Query, and Excel to translate data into decisions.

Beyond analytics, I’ve built a full-stack task manager with React, Node.js, Express, and PostgreSQL, plus an RBAC system with authentication, authorization, and relational SQL design. I’m eager to bring a hypothesis-first, data-driven mindset to product analytics, user engagement, and growth in an early-stage startup.