Venkatesh Devale

I’m an AI/ML Engineer with around 4+ years of experience architecting and deploying scalable machine learning and Generative AI solutions using Python, PyTorch, TensorFlow, and LLMs. I specialize in end-to-end ML pipelines, transformer-based NLP systems, fraud and risk prediction models, and production-grade MLOps workflows.

At Liberty Mutual Insurance, I designed and deployed Generative AI document intelligence pipelines to automate claim entity extraction and contextual risk assessment. I also deployed RAG systems and LLM inference APIs using FastAPI, Docker, Kubernetes, MLflow, and AWS SageMaker—reducing manual document review time by 45%—and I built real-time fraud detection pipelines that improved accuracy by 35% with SHAP explainability.

I’ve also delivered measurable outcomes at Mphasis, including credit default prediction accuracy improvements of 87% through data preprocessing and SMOTE-based class imbalance handling, plus anomaly detection pipelines that reduced fraud investigation effort by 40%. Across both roles, I’ve enforced responsible practices using Microsoft Purview and Data Loss Prevention (DLP), and I collaborate cross-functionally in Agile environments to turn complex business requirements into explainable, high-impact AI.