I developed ROUH, a multimodal AI system for non-invasive embryo viability assessment. It combines embryo image analysis with morphokinetic time-series data.
For ROUH, I built and evaluated EfficientNetB0 and LSTM models, combined their predictions through late fusion, and implemented Grad-CAM and SHAP for model interpretability. I also developed a web-based interface for submitting data and viewing assessment results.
At The Hope International Company, I completed hands-on training in machine learning, data analysis, and AI application development using Python. I worked through practical workflows from dataset preparation and exploratory analysis to model development and evaluation.
As an Assistant AI Trainer at CARDNE- Artificial Intelligence & Agriculture Production, I assisted with introductory AI and data training for adult professionals. I also contributed to data preparation, visualization, and business intelligence tasks at MetaServ - G.R.O.W Internship Program, including developing interactive Power BI dashboards.
