09/30/2026 updated
VS
100 % available
Senior Data Scientist & AI Consultant | Forecasting | Applied ML | Decision Intelligence
Lucknow, India
Only remote
Forecasting Models & Time-Series Analysis
Expertise in a wide range of forecasting models including SARIMAX, ARIMA, Holt-Winters, Exponential Smoothing, VAR, TBATS, Prophet, XGBoost with exogenous variables, and TimesFM. Application of hierarchical forecasting with Daily, Weekly, and Monthly reconciliation, multi-model forecasting, and champion-vs-challenger ensembles.
Machine Learning & Deep Learning
Proficiency in machine learning methods such as XGBoost, Gradient Boosting, Random Forest, and exogenous-driver models, as well as deep learning architectures including LSTM, GRU, TCN, and Transformer-based sequence models for time-series and predictive analytics tasks.
Decision Intelligence & Business Driver Integration
Application of what-if simulation, promotion and event impact modelling, holiday effects, driver analysis, root-cause analysis (RCA), and scenario planning to translate forecast outputs into actionable business decisions. Delivered approximately 10-15% improvement in forecast accuracy through systematic business-driver integration.
Validation & Benchmarking
Use of rolling backtesting, time-series cross-validation, and metrics such as WAPE, MAPE, MAE, RMSE, and Bias for rigorous model evaluation across horizons and segments.
Engineering & Development Stack
Development using Python, SQL, FastAPI, React, scikit-learn, statistical models, and Git/GitHub for end-to-end forecasting pipeline and application deployment.
Foundation Models & Zero/Few-Shot Forecasting
Hands-on experience with TimesFM for zero and few-shot time-series forecasting, as well as benchmarking other time-series foundation models as applicable.
Explainable AI & Root-Cause Analytics
Implementation of SHAP TreeExplainer for global feature importance and prediction-level explanations, Pareto defect ranking, machine-shift heatmaps, operator analysis, and process-window optimisation within applied ML systems.
Statistical Process Control
Addition of SPC control charts, Cp/Cpk process capability analysis, and automated Western Electric rule detection for quality and risk intelligence applications.
Multi-Channel & Regional Forecasting
Forecasting across APAC, AMER, and EMEA regions and multiple customer-engagement channels including Email, WhatsApp, Messaging, Live Chat, Phone, and Salesforce-driven workflows, including product distribution analysis and promotional impact modelling.
Expertise in a wide range of forecasting models including SARIMAX, ARIMA, Holt-Winters, Exponential Smoothing, VAR, TBATS, Prophet, XGBoost with exogenous variables, and TimesFM. Application of hierarchical forecasting with Daily, Weekly, and Monthly reconciliation, multi-model forecasting, and champion-vs-challenger ensembles.
Machine Learning & Deep Learning
Proficiency in machine learning methods such as XGBoost, Gradient Boosting, Random Forest, and exogenous-driver models, as well as deep learning architectures including LSTM, GRU, TCN, and Transformer-based sequence models for time-series and predictive analytics tasks.
Decision Intelligence & Business Driver Integration
Application of what-if simulation, promotion and event impact modelling, holiday effects, driver analysis, root-cause analysis (RCA), and scenario planning to translate forecast outputs into actionable business decisions. Delivered approximately 10-15% improvement in forecast accuracy through systematic business-driver integration.
Validation & Benchmarking
Use of rolling backtesting, time-series cross-validation, and metrics such as WAPE, MAPE, MAE, RMSE, and Bias for rigorous model evaluation across horizons and segments.
Engineering & Development Stack
Development using Python, SQL, FastAPI, React, scikit-learn, statistical models, and Git/GitHub for end-to-end forecasting pipeline and application deployment.
Foundation Models & Zero/Few-Shot Forecasting
Hands-on experience with TimesFM for zero and few-shot time-series forecasting, as well as benchmarking other time-series foundation models as applicable.
Explainable AI & Root-Cause Analytics
Implementation of SHAP TreeExplainer for global feature importance and prediction-level explanations, Pareto defect ranking, machine-shift heatmaps, operator analysis, and process-window optimisation within applied ML systems.
Statistical Process Control
Addition of SPC control charts, Cp/Cpk process capability analysis, and automated Western Electric rule detection for quality and risk intelligence applications.
Multi-Channel & Regional Forecasting
Forecasting across APAC, AMER, and EMEA regions and multiple customer-engagement channels including Email, WhatsApp, Messaging, Live Chat, Phone, and Salesforce-driven workflows, including product distribution analysis and promotional impact modelling.
Project history
Architected a modular hierarchical forecasting platform with Daily, Weekly, and Monthly reconciliation by construction. Implemented seasonality patterns and 6-year backtesting across Naive, Holt-Winters, SARIMA, and XGBoost. Built driver-based forecasting, what-if simulation, and champion-vs-challenger model comparison. Created a future-forecast dashboard combining historical and projected demand, Base/Optimistic/Pessimistic scenarios, reconciliation proof, and daily drill-down. Technology: FastAPI, React, Holt-Winters, XGBoost, TimesFM, Decision Intelligence.
Built a zero-install browser demo with executive KPIs, CSV upload, 4/13/26-week forecasts, confidence intervals, alerts, and downloadable reports. Supports model comparison across SARIMA, XGBoost, Holt-Winters, VAR, ensemble, and TimesFM. Includes event-calendar effects and +/-30% what-if simulation for scenario planning. Technology: HTML, JavaScript, Forecasting Models, Scenario Simulation.
Built an end-to-end PCB assembly defect-prediction system using XGBoost on 12,000 records across 9 process parameters, achieving 81.8% accuracy and AUC 0.885. Implemented SHAP TreeExplainer for global feature importance and prediction-level explanations. Added SPC control charts, Cp/Cpk process capability analysis, and automated Western Electric rule detection. Built root-cause analytics including Pareto defect ranking, machine x shift heatmaps, operator analysis, and process-window optimisation. Deployed as a React + FastAPI application with live prediction, risk gauge, stress-test simulation, and real-time SHAP contributors. Technology: XGBoost, SHAP, SPC, FastAPI, React, scikit-learn, Python.