Hola, I’m Hieu, a Software Engineer

firebase_ai gives a typed Gemini client with Firebase Auth and App Check already in the path. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Client-side Gemini calls with App Check, Auth, and Server Prompt Templates. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Server Prompt Templates keep system prompts and tool definitions out of the binary. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as ...

Fintech apps are regulated products, not just code. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence KYC flows, audit trails, and regional licensing. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Design KYC as a first-class user flow with status states. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the b...

Flags let you ship dark and roll out safely. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Kill switches, gradual rollout, and default-safe flags. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Default-safe means the flag-off path is always a working product. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and cal...

Flags without kill dates become permanent complexity. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Kill dates, owners, and metrics that prove the flag is done. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Require an owner and expiry when the flag is created. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and c...

Bad experiment hygiene ships regressions as wins. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence SRM checks, guardrails, and multi-metric decisions. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Check sample ratio mismatch before reading lift. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary....

AI quality is stochastic. Evals make it shippable. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Golden prompts, regression sets, and release gates. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Keep a versioned prompt + dataset in the repo. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary. ...

EU business terms keep changing. Product and legal must track them together. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Core Technology Commission, storefront web distribution, and compliance. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Payment method choice affects store review and analytics. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller...

Enterprise deals often require MDM hooks. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence App config, kiosk modes, and BYOD realities. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Expose managed app config for policies and URLs. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary. The boundary ...

Your average device is not a $1200 flagship. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Low RAM, high refresh, foldables, and automotive. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Design for 2–3GB RAM Android and 90–120Hz phones. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary. The ...

Student data rules are strict and parent-facing. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence FERPA-ish duties, parental consent, and ad bans. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model No ad SDKs in many school deployments. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary. The boundary...

Flutter is not always right. Platform-only APIs, team skills, and UX purity matter. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence A decision framework for teams that feel FOMO from both ecosystems. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Hybrid (native shell + Flutter modules) is a valid middle path. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A co...

Networks fail. Offline-first treats the local DB as truth and syncs opportunistically. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Local source of truth, conflict rules, and WorkManager handoff. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model User-visible queues beat silent data loss. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case v...

Dot shorthands cut repetition when writing enum values and consts in UI trees. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Why the new Dart syntax makes Flutter UI code less noisy. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Less noise in builders means easier review diffs. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validate...

Docs rot when no one owns them. Assign owners with the code. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence ADRs, runbooks, and README ownership. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model One ADR per irreversible decision. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundary. The boundary re...

Most production fires are leaks or chatty APIs. DevTools finds both. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Finding leaked images and chatty APIs in ten minutes. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Snapshot memory before and after a navigation loop. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent...

Desktop apps are multi-surface. Window APIs are catching up with Canonical. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Multi-window state, focus fixes, and native handle access. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Persist window size/position yourself — the framework will not. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use c...

Dev/staging/prod builds are no longer mobile-only. 3.47 brings flavors to Windows and Linux. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Per-flavor assets and —flavor builds join Android/iOS on desktop. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model CI can produce side-by-side installers with different IDs and icons. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or ...

Tokens keep brand consistent when UI stacks differ. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence One source of truth for color, type, and spacing. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Generate per-platform token files from one schema. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the boundar...

Design systems fail from either chaos or dictatorship. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Contribution model, versioning, and deprecation. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Publish contribution RFCs and a deprecation calendar. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the bo...

Desktop mockups hide the hardest mobile problems. This article turns that observation into a small implementation model you can test, measure, and keep boring when the app grows. The problem in one sentence Keyboard, safe areas, small screens, and offline states. The useful question is not whether the API or pattern looks elegant in isolation. It is where state lives, which boundary owns failure, and how a user recovers when the happy path disappears. A practical model Review with keyboard open and text scale 1.3 minimum. Start with one explicit owner for the behavior. Keep widgets responsible for rendering and user intent; keep IO, persistence, permissions, and retries behind a small interface. That gives you a seam for a fake in tests and a place to record the facts that matter in production. For a Flutter app, the boundary usually looks like this: The widget emits an intent such as load, submit, refresh, or retry. A controller or use case validates the intent and calls the...