Federal digital services — CARM & immigration programs

Challenge

Federal digital programs operate under constraints consumer product teams rarely face: formal software development lifecycle gates, accessibility requirements, security reviews, and the reality that a missed edge case affects real caseworkers and citizens — not just a dashboard metric.

I joined as senior front-end engineer on Deloitte-led delivery for Government of Canada programs, including work on the CARM customs modernization initiative and immigration-related digital services. The work centered on translating complex, manual caseworker and client processes into secure, usable web portals that could pass production promotion bars without rework cycles.

Constraints

• Regulated delivery — Features shipped through government SDLC checkpoints; “move fast and break things” was not an option.
• Accessibility as a gate — WCAG-aligned patterns were part of the definition of done, not a backlog item.
• Dual audiences — Caseworker tools and citizen-facing flows each needed clear error recovery and audit-friendly states.
• Contract continuity — An initial three-month engagement extended multiple times across two programs because delivery kept clearing the next gate.

Approach

I owned complex front-end feature delivery on production portals: typed component boundaries, form-heavy workflows, and integration with backend services exposed over REST.

Engineering practices

• Component architecture — Feature modules isolated validation, loading, and error states so partial API failures did not blank entire screens.
• Accessible forms — Keyboard paths, focus management, and descriptive labels were designed in from the first wireframe, not patched after audit.
• Incremental release — Work was sliced into verifiable increments that could pass security and accessibility review without blocking unrelated features.
• Collaboration with design — Multiple UX recommendations were adopted over original designs when they reduced caseworker steps or clarified citizen messaging.

Working inside government SDLC

Delivery meant aligning engineering choices with program priorities: explicit acceptance criteria, traceable changes, and demos that showed dependability and accessibility under review — not just visual completion.

Outcomes

The engagement delivered production features on national-scale federal digital services with repeated contract extensions — evidence that the work cleared real program bars, not slide-deck demos.

Measurable impact

| Area | Result | | --- | --- | | Programs | CARM customs modernization and immigration digital services (publicly named) | | Role | Senior front-end engineer on Deloitte-led federal delivery | | Tenure pattern | Initial 3-month contract extended multiple times across two programs | | UX adoption | Multiple UX recommendations adopted over original designs |

Metrics are framed to publicly attestable facts. No sensitive program detail, caseload statistics, or non-public system names are disclosed.

What this demonstrates

• Regulated delivery at scale — Shipping under formal SDLC, accessibility, and security expectations.
• Front-end depth on complex domains — Form-heavy, role-based workflows with real operational stakes.
• Trust through extension — Programs renewed the engagement because delivery kept matching evolving priorities.

For the multi-agent engineering system that now accelerates delivery on independent work, see the Agentic OS case study.