harikrishna kandi

I am a Data Scientist with three years of experience building end-to-end machine learning solutions across e-commerce, healthcare, and enterprise automation. I specialize in predictive analytics, NLP, and generative AI, and I focus on turning data into measurable business value.

I design, train, and deploy models using Python, TensorFlow, PyTorch, scikit-learn, XGBoost and Transformers, and I work extensively with vector databases and LangChain for retrieval-augmented generation systems. I have built production APIs and user interfaces, including Flask REST endpoints and Streamlit apps.

My MLOps practice includes CI/CD pipelines with GitHub Actions, containerization with Docker, orchestration with Kubernetes, model tracking with MLflow, and cloud deployments on AWS and Azure. I have experience with Redshift, data warehousing, ETL pipelines, and optimizing systems for latency and scale.

I deliver measurable outcomes — for example, a return-rate forecasting model that achieved 0.8% MAPE, reduced excess stock by 15%, and produced over $500K annual savings. I am collaborative, stakeholder-focused, and driven to build reliable, production-grade AI solutions.