Machine Learning Engineering
Building, training, and deploying models that translate theory into measurable real-world impact. From predictive systems to neural networks, I create AI solutions that perform reliably at scale.
Turning Vision Into Machine Intelligence
Wesley Harrison
Machine Learning Engineer
About Wesley Harrison
As Mark Twain observed, "Loyalty to petrified opinions never yet broke a chain or freed a human soul," a philosophy that drives my work as an AI & Machine Learning Engineer. I specialize in challenging conventional approaches to turn complex data into breakthrough solutions, building neural networks and predictive models that don't just process information, but unlock new possibilities. I thrive where innovation meets execution, designing and deploying AI systems that solve real problems across industries. When I'm not breaking through technical barriers, you'll find me exploring nature with my kids and dog, playing strategic games, or fishing. These activities keep my thinking fresh and often inspire my next project.
Competencies
Driven by the potential of AI to change how things get done, I use SQL, Python, and applied machine learning to turn complexity into clarity and data into decisions. I design and refine intelligent systems that don't just work in theory — they deliver measurable results in the real world. My approach blends technical precision with strategic thinking, bridging the gap between engineering and execution to solve problems that matter and create solutions built to last.
Systems Integration & Automation
Integrating AI into established workflows with precision. I connect data pipelines, APIs, and automation tools so models perform seamlessly in production, not just exist.
Problem-Solving Mindset
Tackling challenges through analytical depth and creative execution. Whether optimizing existing processes or building complete system overhauls, I craft solutions that address both technical requirements and business objectives.
Projects
Predicting Chicago Crime with an Implicit Neural Representation
Developed a spatial-temporal implicit neural representation (SIREN) model to predict Chicago crime patterns, using kernel density estimation for preprocessing to enhance spatial accuracy.
Case Study: Temporal Fusion Transformers
Built and evaluated a Temporal Fusion Transformer (TFT) model for multivariate time-series forecasting, integrating feature engineering and model optimization to improve predictive accuracy over baseline methods.
Case Study: Databricks and Forecasting
Developed and deployed a scalable time-series forecasting pipeline on Databricks, using Delta Lake for efficient data storage and automated workflows for batch and streaming updates. Optimized compute usage to reduce costs while improving model training and inference performance.
FuelFinder
Simulates a Houston-based vehicle's location and, when fuel drops below 10%, clusters nearby gas stations with K-means, calculates distances using the Haversine formula, and recommends the cheapest options within a 10-mile radius.
Explore My
