As a Data Scientist at Minion Labs, I specialize in leveraging advanced data science methodologies to optimize the performance and efficiency of our Minion EMS (Energy Management System) while driving innovation in energy analytics.Project Leadership:⚫ NILM (Non-Intrusive Load Monitoring):-• Led the development of a comprehensive NILM system to monitor industry loads. Managed the project lifecycle from data collection and preprocessing to advanced feature engineering, model validation, and deployment.• Utilized deep neural networks and ensemble methods to achieve robust model performance implemented hyperparameter tuning, with an F1 score of 80% and Recall of 85%.• Tracked model performance using MLflow and managed model versions through a model registry.• Managed data versioning and ensured reproducibility by automating the pipeline with DVC, enhancing scalability and reliability, including automated data uploads to AWS S3.⚫ Advanced Retrieval Augmented Generation Fusion :- Integrated multi-query generation, vector search optimization, and reranking for context-aware outputs. Enhanced retrieval accuracy with custom vector score weighting and reciprocal rank fusion. Utilized OpenAI LLMs and Langchain for advanced language model integration.• Applied prompt engineering techniques, including few-shot learning and Chain-of-Thought (COT) prompting.• Evaluated performance with RAGAS metrics: Faithfulness, Answer Relevancy, Context Precision, and Context Recall.• Monitored and controlled model performance through MLflow and a model registry. Containerized the project with Docker and built a UI with Streamlit.• Containerized and deployed the RAG-Fusion application using Docker, ensuring scalability and easy management.• Automated CI/CD pipeline using GitHub Actions for continuous integration and deployment, pushing Docker images to AWS ECR and deploying to AWS App Runner.