Satyajit Ovelil

Monitored & tuned models forecasting for 2k units/month, developing protocols for long-term accuracy. Customized accuracy metrics to strategically integrate AI principles with business goals. Facilitated stakeholder collaboration to significantly improve and organize accuracy reports. Utilized data-driven tools to measure process performance, check quality & capacity utilization. Oversaw governance standards, streamlined processes, & managed risk logs for consistency.

Led a team of 3-5 Engineers, streamlined AI model life cycle processes, reducing deployment time. Mentored 5+ junior Engineers in ML Frameworks, time series forecasting, git & best practices. Managed refactoring of weather data pipelines for 12+ regions, ensuring consistent processing & reliable workflows. Reduced MAPE by 2% for existing intraday models via rigorous cross-validation & error analysis. Built & deployed ML pipeline for UK electricity price forecasting with cross- functional teams. Developed statistical risk measurement features for energy trading app, deployed securely via FastAPI.

Enhanced forecasting model interpretability across 5+ regions, cutting grid operator's decision time by 28%. Developed modular ML pipelines, boosting efficiency with MLOps in 9 projects & 25+ models. Developed seasonal models for 4 AEMO's NEM region using bias detection for balanced forecasts. Achieved ~95% accuracy in electricity demand forecasting using ensemble models with extensive feature engineering. Prototyped Dagster for scalable weather data pipelines, ensuring reliability in 5+ projects.

Managed audits across 16 branches, boosting compliance and efficiency, and reducing errors. Implemented ML Classifier to predict employee attrition across 3+ departments, 2000+ employees. Created dashboards using Plotly to analyse 2+ yrs of credit cycle data, enhancing decision-making. Optimized large dataset handling (>30 GB) using out-of-core processing for efficient analysis. Developed a Python app to analyse large financial reports, boosting data processing efficiency.