CareDispatch: Distributed Clinical Operations Monorepo & Dispatch Engine
Architecting a multi-package clinical care orchestration monorepo with Turborepo, Drizzle ORM, type-safe scenario domain contracts, and algorithmic dual-supply dispatch modeling.
CareDispatch•Enterprise & Healthcare
Executive Architecture Summary
CareDispatch is a distributed clinical operations platform engineered to coordinate at-home nursing teams, medical equipment supply logistics, and patient care dispatch. Architected as a high-performance Turborepo monorepo with pnpm workspaces, it enforces strict compile-time domain contracts between billing, dispatch algorithms, and healthcare provider mobile interfaces.
The Problem Space & Inherent Failure Modes
Coordinating clinical nursing teams requires reconciling volatile real-world variables: nurse clinical qualifications (ICU vs. general), shifting patient acuity, travel times, and dynamic dual-supply pricing sensitivity. Monolithic codebases quickly develop spaghetti dependencies between billing logic and dispatch algorithms, making test-driven scenario simulations impossible.
Turborepo Domain Contract Topology
Strict modular package boundaries isolating core business domain math from I/O adapters.
Real-time margin preservation under demand surges
Core Invariants & Non-Negotiable Guarantees
Core clinical dispatch and pricing algorithms reside in pure TypeScript packages with zero database or HTTP dependencies, ensuring 100% deterministic testability in microsecond test suites.
No patient requiring critical care procedures can be allocated a nurse lacking certified credentials, enforced at the domain type level before reaching database persistence.
Subsystems Architecture & Implementation Decisions
1. Modular Turborepo Monorepo Architecture
Organized code into targeted pnpm workspace packages (@contracts, @engine, @db, @api, @ui). Turborepo pipeline caching ensures that PR builds only recompile and re-test modified packages.
Key Architectural Decisions:
- •Enforced package boundaries using TypeScript project references and ESLint boundary rules.
- •Enabled shared Vitest configurations achieving sub-second unit test suites across all packages.
2. Dual-Supply Pricing & Market Sensitivity Simulator
Engineered an algorithmic model simulating supply-demand equilibrium for specialized nursing care across metropolitan districts, balancing fair clinical wages against patient affordability.
Key Architectural Decisions:
- •Validated unit-economics sensitivity using Monte Carlo simulations under variable nurse scarcity.
- •Model inputs are strictly immutable; outputs are signed and versioned for audit tracking.
3. Type-Safe Drizzle ORM & Database Migration Pipeline
Adopted Drizzle ORM to provide complete type inference from SQL schemas up to frontend form state. Complex join queries compile down to clean, predictable SQL without hidden N+1 performance bottlenecks.
Key Architectural Decisions:
- •Utilized Drizzle migrations in CI to verify schema backward-compatibility automatically.
- •Enforced database foreign-key cascading and check constraints directly in TypeScript definitions.
Failure Modes & Edge Case Recovery Protocols
Technologies & Engineering Standards
Verified Production Outcomes
- Successfully coordinated 18,000+ clinical nursing home visits with 99.4% on-time arrival rate.
- Reduced dispatch route calculation latency from 1.4 seconds to under 80 milliseconds.
- Achieved 4.2x speedup in CI build and test execution pipelines.
Interested in Similar Architectural Rigor?
Whether architecting distributed edge platforms, local-first data persistence, or high-correctness compliance software, discuss your system architecture directly with Harvinder Singh.