Predictive models, geospatial products, data pipelines, and decision tools for Canada's changing environment.
Production ML pipelines using XGBoost, PyTorch, and LightGBM. Ensemble models, risk scoring, and time-series forecasting.
Multi-tenant platforms with auth, billing, RBAC. Next.js frontends, API gateways, comprehensive automated testing.
Spatial analysis at scale using PostGIS, raster processing, and daily provincial risk surfaces — wildfire at 5 km, flood at 1 km, with fuel measured from 30 m land cover.
Scalable REST APIs with FastAPI and Express. Redis caching, PostgreSQL, automated ETL pipelines, CI/CD.
Environmental risk engine for BC. Ranks wildfire risk daily across 41,857 cells at 5 km, from 27 measured inputs and 180,329 dated ignitions since 1950. A relative ranking for triage, not a probability. Flood susceptibility validated at 0.87 AUC-ROC against observed extent. Fire science and full measurement record open source.
Interactive 3D visualization of the Cascadia Subduction Zone. Real-time seismicity, slab geometry, GPS plate motion. CesiumJS + React.
Real-time salmon migration monitoring for BC. DFO stock data, environmental sensors, interactive maps and time-series dashboards. Next.js + FastAPI + Deck.gl.
AI-accelerated development. We ship production systems in days, not months. Full stack ownership from ML to frontend.
Comprehensive test suites, CI/CD, monitoring, caching. Built to run reliably at scale.
ML pipelines, backend APIs, databases, frontends, infrastructure. One team, end-to-end. No handoff friction.