Shruti Gadewar

I've built and deployed end-to-end AI/ML and data pipelines at Mark and Mary Stevens INI, including a UNet-based computer vision workflow that reduced human quality-review effort by 80%.

I train GAN and diffusion models for anomaly and fraud detection, build validation and interpretability pipelines, and improve LLM relevance through prompt and context engineering. I work closely with domain experts, engineers, and statisticians to turn ambiguous questions into measurable ML solutions.

At the Imaging Genetics Center, I developed production-grade XGBoost and LightGBM classification tools that achieved 93% accuracy, built Python ETL/ELT pipelines, and created RShiny analysis tools for large unstructured datasets. I also applied SHAP to multimodal ensemble classifiers and produced responsible AI documentation.

I've contributed to a Tableau S&P 500 investment dashboard and a serverless AWS task tracker that extracts structured tasks from natural-language input. My work spans Python, cloud platforms, MLOps, generative AI, computer vision, and scalable data workflows.