Selected work

Public sector

Fire-Investigation Case Management Suite

End-to-end modernization of the case-management platform a provincial fire-investigation service relies on daily, replacing a legacy application.

RoleArchitect & Senior DeveloperPeriod2020 — presentWhereOntario public sector
An investigator's worktable: ordered case folders, photographs of burned structures and evidence tags, with a single burgundy thread aligning the photos into one straight line
60+
Investigators served
7
Projects in the suite
~20
Report domains
10
Team at peak

The mandate

A provincial fire-investigation service needed its daily case-management platform rebuilt — investigator case app, data/lookup/security admin app, review app, a shared design-system component library, and a generated EF domain. Quiet software, high stakes: fatalities, serious injuries, exhibits and chain of custody, statements, and financial-loss calculations all live here.

The headline feature is a single comprehensive investigation report assembled from roughly twenty distinct data domains, orchestrated concurrently rather than sequentially.

Architecture leadership

Franklin led the transition from an n-tier monolith to Vertical Slice Architecture: he authored the internal justification paper, ran team-wide training, and designed the foundational structure. Modules like Contacts, Organizations, Persons and Property each bundle their own DTOs, validators, services and repositories.

Cross-module communication runs through a typed mediator with a reflection-based, convention-over-configuration handler resolver — new modules slot in by following the naming pattern, not by editing plumbing. Failures are values (a Result pattern), unwrapped and logged deliberately at the orchestration layer.

The craft details

Server-side PDF/DOCX report generation, document conversion, feature flags, shared-cookie SSO and inactivity timeout — plus AODA/WCAG accessibility, Ontario Design System theming, and bilingual EN/FR delivery throughout.

The suite is also where Franklin's AI-augmented practice shows: a ~460-line assistant design guide encodes team conventions so any AI coding assistant produces standards-compliant code on the first try, paired with a user-feedback automation pipeline.