Chesterfield, Missouri, United States
With over 10 years of data science and machine learning experience, I am a senior data scientist at Bayer Crop Science, where I apply my expertise in statistical analysis, predictive modeling, and causal inference to solve complex problems in customer behavior and new business model design. The focus of my PhD was on computational statistical mechanics and I have strong background in machine learning methods. Some of my recent achievements include developing a probability-of-purchase model for a new product, performing time series forecasting of crop yields, designing geostatistical models of field productivity, and deploying production scale models on cloud platforms. I collaborate with software engineering teams and business stakeholders to deliver data-driven solutions that enhance decision making and optimize outcomes. My current interests revolve around deep learning, causal inference, and graph analysis with applications to business problems and bioinformatics.
- Customer Analytics & Decision Modeling: Leveraged advanced statistical methods to analyze customer data, unveiling key drivers behind purchase decisions. Developed a probability-of-purchase model for a newly launched product leading to key decisions about the product advancement and scaling. - Time Series Analysis & Crop Yield Forecasting: Employed ARIMA and ETS models to perform accurate time series forecasting of crop yields, enabling more accurate risk assessment for outcome-based programs. Integrated findings with business operations to support strategic agricultural decisions. - Geospatial Analysis & Productivity Optimization: Executed complex geospatial queries using PostgreSQL and SQL Server, analyzing agronomic properties and product pricing. Designed and simulated geostatistical models of field productivity, enabling more accurate customer targeting and aiding in resource allocation optimization. - Cloud Deployment & Model Scalability: Collaborated closely with software engineering teams to deploy scalable production models on RSconnect and facilitated data and model sharing through AWS (utilizing EC2, RDS, S3). This effort streamlined software development workflows and reduced time to deployment by more than 50%. - Innovative Deep Learning Applications: Developed a novel deep learning model for tabular data, significantly enhancing the flexibility and integration of predictive analytics in agricultural contexts. - Cross-functional Team Leadership & Collaboration: Acted as a key liaison between data science and software engineering teams, ensuring seamless integration of data-driven models into business applications. Fostered a culture of innovation and data-centric decision making, contributing to the development and launch of 3 new data-driven business models.
- Framework Development for Model Optimization: Designed and implemented a cutting-edge framework for optimizing statistical mechanical system models, significantly enhancing computational efficiency and model accuracy. This framework became a cornerstone for subsequent research and development projects, reducing model optimization times by over 40%. - Innovation in Model Optimization Techniques: Pioneered new, efficient optimization techniques and sampling algorithms, leading to a 30% improvement in computational speed and accuracy for complex models. - Deep Learning for Image Denoising: Applied advanced deep learning algorithms to significantly reduce noise in tomographic images, improving image clarity by 25% and aiding in more accurate analysis of material properties. - Material Modeling for Energy Applications: Constructed detailed atomistic and coarse-grained models for energy storage materials, thin film superconductors, and electrolytes. This research contributed to fundamental research in energy storage material performance and efficiency, which has been awarded the Horizon prize of the Royal Society of Chemistry. - Integration of Experimental and Computational Data: Successfully integrated experimental and computational data to develop effective physical models of porous materials and confined fluids, enhancing the predictive accuracy of material behavior under different conditions. - Microscopic Imaging Data Analysis: Performed comprehensive statistical analysis and reconstruction of microscopic imaging data, leading to improvement in property prediction accuracy for novel materials. - High Performance Computing Software Development: Developed robust software for high performance computing (HPC) environments using MPI and OpenMP protocols, optimizing computational resource use and enabling more complex simulations.
- Developed molecular and coarse-grained models of interfacial energy storage and geochemical systems - Performed interpretation and integration of experimental and computational data for for physical property prediction - Used mathematical models for physiologically based pharmacokinetics (PBPK) - Developed scientific software for molecular dynamics and Monte Carlo simulations.
- Applied theory and simulations to predict fluid properties. - Derived statistical mechanics based equations of state for associating fluids. - Developed software for Monte Carlo simulations.