northend work --inside

See the hard part, not just the interface.

Selected systems shown through their data contracts, safety boundaries, operating controls, and implementation tradeoffs.

northend work --selected

Four systems, inspected at the boundary.

Four repositories. Four different hard parts. Every claim below maps to an implemented boundary, data model, or operating control.

Live product

Adapter system, canonical data model, scoring engine, product and full-stack build

SafeDine

northend inspect safedineSource data structure

Source record

facility_nameinspection_resultviolation_code[]format varies by department

Venue profile

canonical_namelatest_inspectioncomparable_scoreverified structure

Constraint: Health departments publish useful records in incompatible formats, scoring methods, identifiers, and update schedules.
Decision: Normalize the source record before designing the score or the consumer interface.
Result: Comparable profiles across 302,000+ restaurant listings, with scored coverage where official inspection sources are active.

Next.js / TypeScript / Supabase / PostgreSQL / Python / Socrata

Additional engineering

Valuation enginePython · Pydantic · FastAPI · Monte Carlo

ROIC-consistent valuation, reverse solving, scenarios, sensitivities, diagnostics, and a single pure domain core.

Portfolio risk modelPython · Statistical estimation · Local runtime

Beta-weighted exposure, hedge sizing, VaR, stress testing, concentration limits, and estimator-quality checks.

Publication analyticsGA4 API · OAuth · JavaScript · Python

Read-only analytics access and comparative reporting across a multi-publication media organization.

Have a system problem?