Offline-First & Field Application Development
I build applications that work fully without a network and reconcile cleanly when one returns - for field teams, rural operations, and anywhere a dropped connection cannot be allowed to cost a day's work.
Core Capabilities
1. Local-First Data Layers
The local database is the source of truth the user interacts with, not a cache in front of the network. I have built this on Drift and sqflite for Flutter and on Dexie over IndexedDB for the web - reads and writes always resolve locally, so the UI never blocks on a request that may not complete.
2. Background Sync Queues
Mutations made offline are queued durably and replayed in order when connectivity returns, with idempotency keys so a replay does not apply the same change twice. The queue is persisted rather than held in memory, because a field worker closing the app is not a reason to lose a morning's data entry.
3. Conflict Resolution Strategies
Two devices editing the same record while both offline is the normal case, not the edge case. I design the resolution strategy explicitly - server-timestamp authority, field-level merges, or operator review for the records where silently picking a winner is unacceptable.
4. Financial Data Consistency Offline
Payments are the hardest thing to sync, because a double-count is worse than a delay. Reconciling queued local records against payment callbacks received while the device was offline needs idempotency built into the sync layer, not retried hopefully.
5. Native Capability Bridges
Field apps usually need something the framework does not expose - NFC, exact alarm scheduling that survives reboot, background location, hardware peripherals. I build typed bridges between Flutter and native Android so the platform-specific half is contained rather than scattered.
6. Clean Architecture For Testability
Sync logic is where bugs hide and where reproducing a failure matters most. A domain layer with no framework or network dependencies keeps the business rules and the sync behaviour testable in isolation, away from a real device and a real connection.
Why Offline-First Is Its Own Discipline
Adding a cache to an online app does not make it offline-first. The difference is architectural: where the source of truth lives, what happens to a write that cannot reach the server, and who decides the outcome when two offline edits disagree. Those decisions sit at the bottom of the stack, which is why they are expensive to change later.
- Connectivity in the field is not slow - it is intermittent, which breaks different assumptions
- A loading spinner is an acceptable UX on a desk and an unacceptable one on a site visit
- Sync bugs surface as missing or duplicated records, usually noticed days later
- Retrofitting offline support means rewriting the data layer the whole app already depends on
- Teams used to always-on connectivity tend to model offline as an error state rather than a mode
Technologies & Stack
In use across four offline-first applications I have built and maintain:
See It In Practice
Ndoa Plan
Offline-first event coordination platform with a Dexie local store, a background sync queue, and payment-provider callback reconciliation against guest contribution records - operating fully offline during live events.
GIK Souls App
Field follow-up app on a strict clean architecture - pure-Dart domain layer, local-first reads from Drift, and mutations enqueued for a Django backend when connectivity returns.
Sefer
Fully offline Android PDF reader with persistent annotations - custom coordinate mapping over the render pipeline and debounced SQLite writes for rapid library mutations.
Near Field
NFC-dismissed alarm clock where native Android owns exact scheduling, reboot recovery, and the lock-screen ringing flow, bridged to Flutter through typed Pigeon contracts.
How We Can Work Together
Contract
Fixed-scope delivery of an offline-capable app, or an offline data layer added to an existing product.
Consulting
Design review of your sync model and conflict strategy before it reaches production data.
Full-time
Embedded on your team owning the mobile and sync layer long-term.
Based in Nairobi, Kenya - available for remote and on-site engagements across East Africa and globally.
Does your app need to work where the network does not?
Tell me where your users are, what they need to do without a connection, and what cannot be allowed to go wrong when two of them sync at once. I'll come back with a concrete read on the approach.
Get in touch