Profile

---

• Experienced Data Scientist with degrees in mathematics and statistics and a proven track record in deploying machine learning and statistical models to production to solve complex problems in industry.
• Skilled in Python, C++, R, SQL, Git, and Google Cloud Platform (GCP) with significant experience in natural language processing (NLP) and understanding (NLU), generative AI, large language models (LLMs), time series analysis and forecasting.
• Open source examples of my work, including full Bayesian inference for Gaussian process models, time series analysis and forecasting using dynamic linear models, and an R package for variational inference for Dirichlet process mixture models, can be found on my GitHub: https://github.com/michaelellis003.

Experience

---

• A member of The Home Depot’s conversational AI team where we use generative AI, LLMs and other statistical and machine learning techniques to optimize The Home Depot’s contact center experience.

• Directly responsible for building statistical and machine learning models to analyze, predict and quantify uncertainty of energy consumption, weather and climate, with the goal of helping senior business leaders to improve decision making.
• Designed large-scale time-series model to produce forecasts of natural gas demand for 1.2M customers in 8 states improving accuracy by 14%.
• Partnered with NOAA to ingest over 2.2 billion weather data points to produce climate and weather models.
• Developed several R-Shiny web applications. Including an application to automatically calculate the optimal climate normal for each NOAA weather station in the Black Hills Energy’s service territory.

Teaching

---

• Course covering an introduction to probability theory, hypothesis testing and regression analysis.
• Taught 2 course sections of approximately 40 students each semester.

• Taught 3 course sections of approximately 30 students each semester.

Education

---