At Impact Analytics, I built and tuned forecasting models for enterprise clients, processing 20M+ rows and selecting models by segment. The pipeline used time-series features and Ridge regression, XGBoost, and LightGBM ensembles.
I designed and deployed an agentic AI forecast-monitoring system that detects anomalies, investigates root causes, and sends plain-language alerts. It reduced tickets routed to the team by 30%.
I also built a RAG knowledge assistant using embeddings and FAISS to answer questions grounded in client documentation, SQL scripts, and EDA references.
At Vedanta Ltd – Cairn Oil & Gas, I built an XGBoost predictive-maintenance model using SCADA sensor data to predict equipment failure within a 1,000-hour horizon. I also built Power BI dashboards for plant and site health, with anomaly detection and variance analysis.

