Santhosh Kumar Selvam

o Led AI/ML implementations and delivery of o9 AI-powered Enterprise Planning solutions for large Retail, Distribution, and Consumer Products clients, including Chow Sang Sang, Pandora, Mondelez, Starbucks, T-Mobile, Coty, and Lego using a 10-member team.
o Served as an advisor to C-Level executives, helping them leverage data assets for better decision-making.
o Liaised with R&D and Product Teams to develop core data science building blocks.
o Developed standardized project plans and sprint reviews to minimize production escalations.
o Designed custom AI/ML solutions for complex projects that required extending standard product functionalities.
o Led debugging and performance improvements in big data environments, optimizing large-scale AI solutions.
o Built Proof of Concept multi-agent AI systems that orchestrate multiple LLMs to optimize enterprise workflows.
o Developed AI-driven automated data quality checks and exploratory data analysis (EDA) using LLM-based multi-agent AI systems for detecting data quality issues.
o Created Agentic AI systems for improving product documentation, enabling automated summarization, knowledge extraction, and contextual enhancements from disparate data sources.

o Developed and deployed a Multivariate Multistep Global-to-Local Forecasting Engine using LightGBM and Optuna for Philips Personal Health in Europe (including Amazon accounts totaling €600mn revenue).
o Built a Forecast Ensembling Engine for combining forecasts from statistical models, tree-based algorithms (XGBoost, LightGBM, H2O.ai AutoML), and neural networks (TensorFlow).
o Designed a demand sensing solution for weekly demand signal generation in North America.
o Created Shapley-value-based explainability artifacts to improve forecast interpretability.
o Developed forecast classification based on confidence levels to improve adoption among planners.
o Led forecast automation and standardization, reducing turnaround time for forecast generation, validation, and KPI measurement using a team of Data Scientists for Philips Personal Health Supply Chain

o Designed Hierarchical Time Series models and Advanced Forecast Disaggregation methods to improve SKU-location level demand forecast accuracy.
o Created a Time Series Pattern Recognition and Clustering algorithm for Building Climate Control Simulation Studies using IoT thermostat data.
o Designed and built a Product Recommendation & Substitution Tool, identifying similar SKUs for inventory optimization and SKU rationalization.
o Created multi-year daily sales forecasts and promotions analysis for an e-commerce business.
o Built a tool for detecting anomalous financial transactions on corporate credit cards for financial auditing.
o Built an R Shiny application for Reliability & Warranty Analytics, replacing the industry-standard Weibull++ tool.

• Graduate of Caterpillar's intensive 18 months rotational Leadership & Technical Development Program
• Worked on Supply Chain and Manufacturing Data Analytics
• First to develop and deploy data driven process management in the BCP Paint Shop and Assembly Line