Mani Krishna M.

Data Scientist @Amygda

Derby, GB

Designed and implemented four production-grade machine learning models to predict failures in baggage scanning machines and trains, achieving a 20% reduction in unplanned downtime and enabling proactive maintenance strategies.• Built a scalable ETL and prediction pipeline using Python and PostgreSQL, optimized SQL queries to reduce pipeline runtime by 60%, significantly improving batch processing efficiency and reducing operational costs• Utilized PySpark on Google Cloud Dataproc to process and analyze files, enabling efficient handling of large-scale datasets.• Established robust MLOps workflows using MLflow for experiment tracking, GitHub for version control, and Docker for containerization. Deployed models on Google Cloud, achieving 99.5% system reliability through comprehensive unit testing.• Resolved production bugs in machine learning models, integrating real-time feedback loops to improve prediction recall by 15%.• Developed an XGBoost-based time-series model for train fault prediction, achieving a 0.73 recall. Delivered insights via a Streamlit UI application and deployed the solution using Docker and Google Cloud Run as part of a successful proof of concept (PoC).• Collaborated with product teams to integrate models into production and effectively communicated technical insights to clients through clear and concise updates• Enhanced an internal data science tool by adding anomaly detection and data analysis, improving interpretability and diagnostics.• Built a knowledge graph using GPT-4 and Neo4j for unstructured documents, increasing retrieval efficiency by 25%.

Gayatri Vidya Parishad College of Engineering (Autonomous)

Bachelor of Technology - BTech, Computer science and Engineering

University of Leeds

Master of Science - MS, Data Science and Analytics