Yashika Basapure

I’m a junior engineer focused on deep learning, Bayesian inference, geospatial pipelines, and production ML systems. In my current role as a Data Scientist Jr., I’ve built end-to-end geospatial data pipelines for satellite imagery and raster/vector datasets—covering preprocessing, geo-referencing, and transformations with attention to data quality at scale. I apply LLMs and LVMs to multimodal geospatial tasks (including segmentation with SAMGeo) and deploy ML pipelines for scalable training and inference on Microsoft Azure.

I also bring a research mindset to applied problems. I engineered a modular virtual brain inference pipeline that integrates diffusion MRI connectomes with neural mass models for GPU-accelerated brain simulation, and I implemented Bayesian inference for uncertainty-quantified virtual brain twins for clinical seizure prediction. Across projects, I’ve delivered practical impact—building a fraud detection system with explainability (SHAP/LIME) on imbalanced data and fine-tuning Whisper for conversational Hindi ASR using an end-to-end data pipeline and fair lattice-based WER evaluation.