Amazon SageMaker
Amazon SageMaker is a platform for building, training, and deploying machine learning models.
What It Is
Machine Learning is A subset of AI focused on systems that learn from data. On this site, it matters because it transfers across technical, operational, and venture work instead of staying trapped in one narrow context.
Learn more: https://en.wikipedia.org/wiki/Machine_learning
Curiosities and Interests
This page collects the questions, patterns, and topics that repeatedly draw my attention.
Demographic Prediction Model and Machine Learning Pipeline
Built predictive audience demographic datasets using graph databases and AI/ML. Trained models to predict demographic attributes of mobile device audiences...
Enhanced IP Geolocation Model for Security and Verification
Designed a novel IP geolocation confidence model combining location observation data with network topography analysis to improve the accuracy of geographic...
EPA SBIR Phase 1 — Residential Energy Optimization Algorithms
Won a $66,890 EPA SBIR grant to develop adaptive energy management algorithms using sensors and machine learning for residential HVAC optimization — an early...
Location Observation Fraud Outlier Identification
Used Tableau and sampled data to identify impossible or improbable movement patterns in location observation datasets, then designed a system to tag and filter...
Machine Learning and Data Science
Projects involving predictive modeling, ML pipelines, feature engineering, and data-driven decision systems.
Scoring Algorithm for Customer Revisit Attribution
Designed, implemented, and tested 150 iterations of a scoring algorithm to determine the likelihood of a customer revisiting a physical store, based on mobile...
Similarity & Clustering Systems
Systems measure likeness between entities using features and distance metrics so they can drive insights, grouping, and product behavior.
Voronoi Regions for International Geofencing
Created an algorithm to build an optimal international geofencing system for hundreds of millions of postal address regions, incorporating feedback from 32...