Independent ML prediction platform and AI consultancy. Built and deployed AURA, a census-tract-level hurricane impact model covering 5,668 Gulf Coast tracts across 14 storms (R² = 0.924). Translates complex data problems into accessible, production-grade tools.
Overview
Prerogative AI is an independent ML prediction platform and AI consultancy founded by Liam Guest. PAI sits at the intersection of rigorous statistical modeling and product-grade execution — building tools that are both technically defensible and usable by real people. The platform hosts live predictive models and takes on consulting engagements that translate complex data problems into accessible, actionable outputs.
Live Models
AURA — Hurricane Impact Prediction Model
AURA is PAI's first model. It predicts hurricane impact at the census-tract level across the Gulf Coast, covering 5,668 tracts and 14 historical storms.
- R² = 0.924 — strong predictive performance across diverse storm conditions
- Inputs include storm track data, wind speed, tract-level demographics, and geographic exposure
- Outputs are tract-level impact scores applicable to emergency resource allocation, insurance risk modeling, and post-storm response planning
- Live technical showcase at prerogativeai.com/models/aura
Current state: The AURA page presents the model's methodology, performance, and results. It is a technical showcase — not yet an interactive tool. The roadmap includes API-gated access to continuously updated model outputs, with tiered access for researchers, insurers, and emergency planners.
Interested in early access or a consulting engagement? → liamguest71@gmail.com
Kalshi Market Forecasting (coming soon)
A prediction market forecasting model targeting Kalshi contracts. In development.
Consulting
PAI takes on engagements at the intersection of ML, biostatistics, and product. Work spans model development, analytical infrastructure, and translating data outputs into stakeholder-facing tools. Background in M.S.P.H. Biostatistics & Data Science (Tulane University).
Stack
- Python — modeling and data pipelines
- React / Vite — platform frontend
- Vercel — deployment
- Anthropic API — LLM integrations