AI/ML Engineer based in Bengaluru. I build production ML systems that help B2B SaaS companies understand their revenue pipeline before the quarter closes.
Four years of shipping: forecasting, propensity modeling, marketing mix, pipeline projection, model monitoring. 25+ systems in production across real customers, real data, and real consequences when a model drifts.
revenue forecasting EOQ booking prediction from CRM signals. Multi-horizon. Leakage-safe cross-validation. Macro adjustment layers that keep overall pipeline MAPE below 10%.
pipeline projection 8+ ML model families (propensity, deal size, stage transitions, demand gen) aggregating bottom-up into daily multi-quarter projections. Millions of records scored daily. 15+ enterprise deployments. Pipeline F1 at 80%+, booking conversion at 85-95%.
marketing mix modeling Measures each channel's pipeline contribution via Hill saturation curves, adstock decay, and seasonal decomposition. Scenario planner for budget reallocation. 5-15% MAPE across 60+ channels.
propensity scoring Four-model suites across accounts, leads, opportunities, and demand gen. Daily conversion likelihood scores, SHAP-based explanations, statistical fallback for thin-data customers.
ML frameworks Generic regressor baseline with RandomizedSearchCV tuning, quarter-aware validation, and production scoring/writeback. Model metric dashboard tracking MAPE/wMAPE/MAE/RMSE across all deployed models for degradation detection.
Python XGBoost LightGBM CatBoost scikit-learn SHAP
PySpark BigQuery GCP Dataproc Airflow SARIMAX SQL