Hello! I'm Michael, a data scientist with a math and statistics background who likes to build things. My work spans Bayesian methods and causal inference to LLMs, GenAI, and AI agents. I believe good data science requires good software engineering. Currently, I'm especially interested in the intersection of causal inference and agentic AI, building systems that use causal reasoning for counterfactual simulation and intervention planning.
The methods and systems I work on most often
Building and deploying ML systems that work in production.
Large Language Models · Generative AI · NLP & Voice Interfaces · Deep Learning · Reinforcement Learning · Forecasting & Predictive Modeling · MLOps & Model Serving
Using statistical models and experiments to estimate effects and make decisions under uncertainty.
Bayesian Inference · Causal Inference · A/B Testing & Experimentation · Time Series and Spatial Datasets · Monte Carlo Methods · High-Dimensional Probability & Statistics
Writing maintainable software and the tests, automation, and infrastructure needed to run it.
Python · C++ · TypeScript & JavaScript · Next.js & React · Node.js · System Design & Architecture · CI/CD & DevOps · Cloud Infrastructure (GCP)
Libraries and tools for statistical computing, ML, and Python development
Sequential Monte Carlo and particle filtering in JAX
bayesian-inference · hidden-markov-models · jax · particle-filter · python · sequential-monte-carlo · state-space-models
Language model research platform for Apple Silicon — 24 architectures, MLX-native training, LoRA/DPO/GRPO, and experiment tracking
apple-silicon · deep-learning · language-models · llm · lora · machine-learning · mamba · mixture-of-experts · mlx · python · rlhf · transformer
Production-ready Python package template with uv, ruff, pyright, and GitHub Actions CI/CD
ci-cd · github-actions · pyright · python · ruff · template · uv
Bayesian Gaussian mixture models with Gibbs sampling and variational inference in R
bayesian-inference · gaussian-mixture-models · gibbs-sampler · mixture-models · r · variational-inference
Bayesian inference for Dirichlet process mixture models using Gibbs sampling and variational approximations in R
bayesian-inference · dirichlet-process · gibbs-sampler · mixture-models · nonparametric-statistics · r · variational-inference
Variational Bayes and CAVI algorithms from Ormerod & Wand — approximate Bayesian inference examples in R
approximate-inference · bayesian-inference · machine-learning · r · statistics · variational-inference
Full Bayesian inference for Gaussian process spatial models using MCMC in R
bayesian-inference · gaussian-processes · geostatistics · mcmc · r · spatial-statistics
My work in data science and background in mathematics and statistics
Walmart
· Current
Led development of a production causal inference library identifying ~$100M in incremental revenue opportunity through optimized sales-meeting targeting for Walmart Marketplace partners.
The Home Depot
Built text and voice customer-support systems with generative AI, large language models, NLP, and cross-functional product teams.
The Home Depot
University of Arkansas
Black Hills Energy
Worked on a large-scale forecasting system for natural gas consumption.
University of Arkansas
Master's thesis: Sequential Monte Carlo Methods for Hidden Markov Models
University of Arkansas