Solutions-oriented full-stack Machine Learning Engineer, Data Scientist, and AI Developer, with 12+ years of experience designing high-performance distributed systems and applications. Proven track record of leveraging modern Machine Learning, Data Science, and and AI frameworks, containerization technologies, and serverless architectures to solve complex business challenges. Passionate about clean code, structured output, performance optimization, and delivering exceptional user experiences that drive measurable outcomes.

• Data Science: Polars, Pandas, NumPy, SciPy, Scikit-Learn
• MLOps / LLMOps / AI: PyTorch, HuggingFace, Ollama
• Cloud Infrastructure: Google Cloud Platform, AWS
• DevOps: Docker, Terraform, Kubernetes
• Frameworks: Streamlit, Nginx, Flask
• Programming: Python, R, C++

• Designed and deployed an enrollment forecast model using RandomRiver distributed in a production MLOps pipeline; returning a forecast of total enrollments on a by-day, by-school resolution; achieving 95%+ accuracy for each of the three years it was used for guiding business outcomes
• Independently developed and delivered a full-stack AI chatbot using HuggingFace + Ollama + FastGCP featuring multiple agents, including a SQL agent to query data from our EDW; integrated with a custom MCP server; served using a Streamlit front end; deployed on GCP with IAP behind a Google Cloud Load Balancer to protect PII data; enabling leadership to leverage data inference for driving business decisions
• Sole developer for a propensity model using PyTorch, trained on Google Analytics activity data and returning a Propensity Score for the likelihood that a customer will apply and/or enroll in our program; deployed and automated on GCP using Cloud Run; allowing for a measurable improvement in application nurturing
• Engineered a scalable LLMOps pipeline using llama.cpp + LoRA for distributed training of custom transformer models on multiple GPUs; including quantization and conversion to GGUF format for Ollama ingestion; reducing quality loss to 5-15% performance degredation and achieving 50-75% memory saving, resulting in lower dropout and optimal alpha, usually 2x rank value
• Developed a Media Mix Modeling (MMM) pipeline using pymc-marketing and Meridian to analyze spend allocation among 400+ individual marketing channels spanning 30+ individual states; delivering model evaluation insights, demand assessment, and budget optimization for our marketing department
• Engineered a RAG protocol using Phi-4 + Ollama, integrating vector search and LLM inference; streamlining knowledge extraction, analysis, and classification of agent-client conversations; enabling an estimated 75% faster review workflow of our dataset of 200K+ transcribed conversations
• Developed an A/B Testing protocol using a Bayesian framework with Monte-Carlo sampling to optimize campaign cadence based on application-to-enrollment conversions; increasing marketing spend allocation efficiency by an estimated 50%
• Designed and implemented end-to-end pipelines for all production API frameworks and connections to third-party platforms such as Google and Meta with an automated CI/CD protocol using GCP Source and Cloud Run; managed IAM authorizations for our department as Admin of our GCP project
• Spearheaded and implemented department-wide version control using git and Google Source Code