Pádraig Lucey
Veterinarian and dairy nutrition researcher. MVB, MPVM, Ph.D., DACVPM, DACVPM-Epi
I am a veterinarian from Ireland living and working in the dairy industry of California.
I am building integrated data layers and agent loops for dairy consulting.
My work and research interests include population health, nutrition modeling, machine learning and data science.
Writing and talks
- Mining dairy data with AI agentsTalk · California ARPAS CEC 2026 PDF
- Precision feeding dairy cow pens with machine learningTalk · California ARPAS 2024, Coalinga
- Predicting dairy pen DMI with machine learningTalk · California ARPAS 2023 (student)
- Predicting dairy pen dry matter intake distribution using machine learningTalk · ModNut 2022
- Generate tabsets in a for loop with R Markdown
- ANOVA comparison of datasets
- Overlaid plots in ggplot2
Tools for AI agents
Agents can't do this arithmetic reliably in their heads. These give them the methods that can.
Ration Lab
Fully featured ration formulation software built in Rust, with multiple linear and nonlinear solvers. Includes chance-constrained solutions based on the variation in feed tests, and the variation between cows in a pen from my PhD thesis. Available online, or as an MCP server your AI agent can connect to and run.
- In your browser
- lab.dairydatascience.com
- MCP server
https://lab.dairydatascience.com/mcp- Claude
- Add a custom connector with the address above
Dairy Data Tools
A dairy's own DairyComp and EZfeed exports, read by your AI assistant: the herd, each pen and cow, pen cows joined to pen feed, a weekly or visit review, and a ration change with the season taken out. Worked out by tested scripts, not guessed.
- Claude
- Customize > Plugins > Add marketplace: padge01/dairy-data-tools
- Codex and others
npx skills add padge01/dairy-data-tools- On your computer
pip install dairy-data-tools
BESTPRED
USDA's best prediction of dairy lactation records, as tools any AI agent can call: 305-day and mature-equivalent milk, fat and protein, reliability and persistency from a herd's or a cow's test days. A Python port checked against USDA's own program.
- Claude
- In Claude's plugin directory: BESTPRED
- Any agent
https://bestpred.dairydatascience.com/mcp- Codex
codex plugin marketplace add padge01/bestpred-python- On your computer
uvx bestpred-mcp
Papers
Predicting the distribution and dry matter intake of cow pens with machine learning New Lucey, Lean, Stepanchenko, Martins, Hristov, Rossow · Animals, 2026
Feeding preweaning Holstein calves a synbiotic supplement increases their energy-corrected milk yield as lactating cows Journal of Dairy Science, 2026
Describing the distribution type of DM intake for dairy cow pens based on pen characteristics animal, 2023
Evaluation of cow-side meters to determine somatic cell count in individual cow quarter and bulk-tank milk samples Animals, 2023
Effects of mannan-oligosaccharide and Bacillus subtilis supplementation to preweaning Holstein dairy heifers on body weight gain, diarrhea, and shedding of fecal pathogens Journal of Dairy Science, 2021
Projects
- EKS calculators for UpLookBuilt the Bovaer and Rumensin calculators for Elanco; advised on the cow-day and manure calculators.
Code
- Commits
- 16,063
- Pull requests
- 2,507
- Reviews
- 3
- Issues
- 28
- dairy-data-toolsSkills for dairy nutritionists and their AI agents: read a farm's DairyComp and EZfeed exports. From Dairy Data Science. · Python
- bestpred-pythonPython translation of USDA's BESTPRED lactation records (305-day yields, persistency, daily curve). Same numbers as the original. · Python
- xteink-pokemon-gameAn unofficial reading-powered Pokemon companion game for the Xteink X3, built on CrossInk. · C
- homelabA battery-powered, terminal-only wedge computer for teaching kids to code at home: Raspberry Pi 5, mechanical keyboard, thermal printer. · HTML