CAT Used Price Advisor
A dealer-facing tool, powered by machine learning models, that prices more than 70,000 pieces of equipment worth over $12B.
Portfolio
A handful of projects I'm proud of, spanning equipment pricing, genomics, and plant breeding — the kind of work I keep coming back to when explaining what I actually do all day.
Hedonic Price Index
An index that separates real market price movement from quality differences, used by CAT Financial and senior leadership to sharpen pricing strategy across sale channels.
Demand Forecasting Model
A forecasting model that predicts industry size five quarters out, using price-trend data from auction and retail markets along with consumer behavior.
AI-Assisted Engineering
AI-assisted engineering workflows spanning code generation, debugging, refactoring, and rapid prototyping, accelerating development and shortening delivery cycles.
Executive Brief Generator
An LLM-powered workflow that turns analytical outputs into one-page executive briefs, accelerating insight communication and decision-making.
Executive KPI Dashboards
KPI dashboards and decision support for SVPs, translating complex analyses into recommendations for product strategy and patent disputes.
GenCoCo Tool
An interpolation-based genomic coordinate conversion algorithm, served via an R Shiny app, that cuts conversion time by 90%.
Cost Savings
Queryable BLAST databases built from public genomes, paired with a haplotype-based method for identifying gene-editing targets with greater statistical power — avoiding $750K in cost while widening research accessibility.
Parent-Progeny-Trio QC
A parent-progeny-trio QC framework that delivers 100% data quality control with zero sample mix-ups.
Genomic ML Pipelines
Production pipelines processing 3TB of genomic data a day, cutting manual effort by 70% and making downstream analytics faster and more reliable.
Unified Feature Repository
A genomic feature repository in BigQuery that became the team's single source of truth for crop analytics, and the foundation for everything built on top of it since.
Imputation Algorithm
A proprietary imputation method that took usable data return from 30% to 92% per feature — a meaningful jump in how much of the data could actually be put to work.
Gaussian Mixture Modeling
Gaussian Mixture Models applied to improve true variant detection by 70% in genomic datasets, strengthening the data quality everything downstream depended on.
Trait Discovery & Breeding
Insect-resistance genes mapped using mixed-model regression, then deployed into wheat varieties through hypothesis-driven experiments — resulting in products that use less pesticide and support more sustainable farming.
Genomics & Bioinformatics
Sequencing-based bioinformatics methods to detect genome deletions and identify duplicate entries in public gene banks, cutting detection time by 60% and curation costs by 50%.
Novel Popgen Discovery
Principal component analysis and hierarchical clustering applied to identify a novel wheat population genetic group, contributing a new source of genetic variation to the research community.