End-to-end work in SAP automation, data engineering, and applied statistics — built to solve real supply chain problems.
Supply Chain Transformation
Enterprise SAP automation platform for global manufacturing operations
The Central Parameters organization historically executed critical SAP master data and procurement activities manually. Every manufacturing unit had slightly different business rules, making standardization difficult and requiring analysts to spend thousands of hours on repetitive data entry instead of higher-value supply chain work.
Designed and developed a Python-based automation platform that orchestrates end-to-end SAP transactions across multiple enterprise data sources. Requests are extracted from internal data products, SAP systems, and the enterprise data lake, then validated, standardized, and executed according to plant- and region-specific business rules. Depending on the process, automations interact with SAP through GUI scripting or SAP S/4 APIs, allowing reliable execution while preserving business controls.
Automates high-volume operational processes including material master part number extensions, purchase order creation, contract creation, pricing updates, and other master data maintenance activities. Each workflow applies plant-specific and regional business rules while maintaining standardized execution across the organization.
2
Enterprise data integration
Combines data from internal data products, SAP, and the enterprise data lake. Information is extracted, cleaned, validated, enriched, and transformed into a standardized format before execution, ensuring consistent results across business units.
3
Hybrid SAP execution
Processes are executed through the most appropriate interface: SAP GUI automation for complex transactions and SAP S/4 APIs where available. This hybrid approach balances reliability, speed, and compatibility with existing enterprise systems.
4
Production-grade engineering
Source code is maintained in GitHub using pull requests, code reviews, CI/CD pipelines, dependency scanning, and security analysis. Deployments are automated to a dedicated Windows Server environment that executes production workloads for the supply chain organization.
Machine Learning / Supply Chain
Machine learning engine for SAP MRP group classification
Each manufacturing plant maintained its own decision tree to determine the correct MRP Group when extending materials into SAP. Although the underlying inputs were nearly identical—such as material group, part type, and part prefix—every plant documented its logic independently in large PDF decision trees. These documents were difficult to maintain, time-consuming to train new analysts on, and frequently became inconsistent across manufacturing locations.
Designed and deployed a machine learning classification engine that predicts the appropriate MRP Group directly from material attributes, eliminating the need for manually maintained decision trees and hard-coded business logic. Built a production dataset from existing SAP material master records across multiple manufacturing plants and trained a CatBoost multiclass classifier capable of learning plant-specific classification behavior while leveraging shared patterns across the enterprise.
Created a consolidated training dataset from production SAP material master data, combining features such as plant, material group, part type, prefix, COMAR code, term code, and additional master data attributes used during material creation.
2
Feature engineering & model training
Evaluated multiple classification approaches before selecting CatBoost for its native handling of categorical variables and superior multiclass performance. The model learned both enterprise-wide classification patterns and plant-specific business rules without requiring manually encoded decision trees.
3
Production inference
Integrated the model into the material extension automation platform, allowing new materials to be automatically classified during SAP view extension. The prediction engine replaced extensive if/else logic and eliminated the maintenance of dozens of independently managed decision tree documents.
Data Science / Time Series
Residential electricity consumption forecasting
Residential energy consumption exhibits strong seasonality, long-term trends, and irregular fluctuations, making accurate medium-term forecasting challenging. Reliable forecasts require careful preprocessing, statistical validation, and model selection.
Developed a complete forecasting pipeline as part of my M.Sc. thesis at Universidad Autónoma de Coahuila. Historical household electricity consumption data was cleaned, aggregated to daily observations, analyzed for trend and seasonality, and modeled using both statistical and machine learning approaches. Model performance was evaluated using multiple forecasting metrics to identify the best-performing methodology.
Processed raw smart meter readings into a clean daily time series through aggregation, missing value treatment, and quality validation to produce a robust modeling dataset.
2
Time series analysis
Applied STL decomposition to isolate trend, seasonality, and residual components. Verified stationarity using Augmented Dickey-Fuller (ADF) and KPSS tests before training forecasting models.
3
Model evaluation
Compared statistical and machine learning forecasting approaches using MAE, MSE, and MAPE, selecting the model with the best predictive performance for 30-day forecasts.
Teaching / Higher Education
Engineering instructor for Data Science, Statistics & Programming
Engineering students often struggle to connect theoretical concepts in statistics, probability, and programming with practical engineering and data science applications. Developing analytical thinking requires more than lectures—it demands real-world examples and hands-on problem solving.
Teach undergraduate engineering courses at Tecmilenio University in an online format, designing lectures, programming exercises, and practical case studies that bridge statistical theory with modern data science and software development practices. Classes emphasize applying quantitative methods to real engineering problems using Python and industry-relevant datasets.
Teach descriptive and inferential statistics, probability theory, hypothesis testing, regression, forecasting fundamentals, and statistical decision-making through practical engineering examples.
2
Programming & data science
Guide students through Python programming, data analysis, and introductory machine learning concepts using real datasets, encouraging reproducible analysis and problem-solving skills.
3
Interactive online instruction
Deliver live virtual lectures, develop course material, evaluate assignments and projects, and mentor undergraduate engineering students while fostering participation and collaborative learning.