Email Ayub

Secure systems. Real workflows. Production ownership.

I build best when
the problem is still messy.

I learn the workflow, make the constraints visible, build the smallest reliable system, and stay for the production reality.

Illustration of Ayub working at a laptop.

Software became the way I improve operations.

I started in retail and warehouse operations, where problems were immediate: a customer needed the right answer, a shelf was empty, a shipment was wrong, or a team needed a clear priority.

At ESDC, that operational thinking moved into secure federal APIs. At Canadian Bank Note, it expanded into testing, country implementations, online passport workflows, payments, and production support.

Recent agency and product work has added TypeScript, React, cloud deployment, internal tools, analytics, and AI-assisted delivery. The tools changed. The working habit did not.

Three rules I keep returning to.

Find the expensive friction.

The loudest request is not always the real problem. I map the workflow before choosing the interface.

Put correctness where it belongs.

Transactions, idempotency, tenant isolation, validation, and tests carry more trust than optimistic copy.

Delivery includes the last mile.

A green build is not production proof. I check data, APIs, deployment behavior, handoff, and failure recovery.

Software is how I improve the operation.

I start with how people work, then choose the right system. AI can research, classify, draft, extract, or route, but models only become useful when software gives them structure, tools, safeguards, and a clear human decision.

  1. Observe

    Watch the real work and find where information, responsibility, or time is being lost.

  2. Map

    Turn the current process, exceptions, users, and acceptance criteria into a visible workflow.

  3. Choose

    Select the smallest useful combination of interface, deterministic code, APIs, models, and people.

  4. Build

    Package the workflow into software people can use repeatedly, not a one-off prompt.

  5. Guard

    Add schemas, validation, permissions, tests, audit history, and human approval where trust matters.

  6. Ship

    Verify the runtime, data, integrations, handoff, and failure recovery, not only the build.

  7. Learn

    Use real outcomes and operator feedback to improve the next version.

ModelProduces a prediction or generated result from an input.

AgentUses a model with a defined responsibility, tools, state, and stopping rules.

WorkflowCoordinates deterministic steps, agents, APIs, and human decisions.

Internal toolPackages a workflow for a team doing recurring work.

Public productAdds the reliability, access, support, and operating boundaries needed for external users.

Public GitHub is only part of the work.

The public profile currently holds 9 repositories. Much of my recent work is in private repositories and internal systems, so the public contribution graph is not a complete picture.

The graph shown here is a private-inclusive screenshot supplied by Ayub and captured September 3, 2026. It is not independently verifiable from the public profile.Open GitHub profile
1,630 GitHub contributions in the last year, including private activity, captured September 3, 2026.

1,630 contributions in the last year, including private repositories.

MasjidPassDartBasicCalculatorJavaOthelloJavaSimonC++

Curious, practical, and happiest near the real problem.

Outside the work, I keep learning through product experiments, systems research, and building tools for the people around me.

Email Ayub