ahsan.dev Résumé

Available for senior mobile roles and contract engagements · Top Rated Plus on Upwork

Senior mobile engineer Android · Kotlin · Compose

Ahsan Islam — nine years of production Android delivery in Kotlin. Specialised in white-label commerce platforms, payment gateway integration, BLE hardware companions, wearables and production computer vision, delivered for teams in the UAE, Saudi Arabia and Pakistan.

Check role requirements Selected work Book a 30-min call Contact
0+ yrs shipping
0k+ Play installs
0 payment gateways
0nd · AWS AI/ML

Shipped and live on Google Play

01 — About

Profile

Portrait of Ahsan Islam
Lahore, Pakistan · Remote Working with Abu Dhabi, UAE English · Urdu · Punjabi

Nine years delivering mobile products from concept to production at scale, Android-first, across phones, wearables and connected BLE hardware. Currently a senior Android developer at COFE App (Abu Dhabi), engineering a white-label commerce platform in which a single configuration produces a fully branded application.

Previously responsible for Android delivery on fintech and insurtech products: integrated five bank and mobile-wallet payment gateways behind a single checkout, raising transaction success by 50%, and delivered health products synchronised with four health and wearable platforms.

Primary technical focus is architecture: Clean Architecture with strict layer boundaries, SOLID applied at the seams — interfaces at module boundaries, dependencies inverted onto abstractions, one reason to change per module — together with testability, performance profiling, ANR and crash reduction, and CI that releases only on green.

New platforms are adopted in production rather than in training: a first Kotlin Multiplatform project reached a published Play release running one Compose codebase on Android, iOS and Wasm web. Also led the team that placed 2nd of 25 companies in the AWS and Epiphany AI/ML Startup Reactor competition with a vehicle-damage computer vision model.

30%faster load times via server-driven UI
80%better recognition accuracy on the AI feature
25%sensor accuracy gain on BLE tyre pressure
02 — Experience

Professional experience

  1. Senior Android Developer

    Sep 2023 — Present

    COFE App · Abu Dhabi, UAE (remote)

    • Deliver 12+ brand apps from one Kotlin/Compose codebase using per-brand vendor configurators on a shared vendor-app framework.
    • Own the payment layer: two gateway SDKs behind one provider abstraction — Tap Payments GoSell and Checkout.com Frames — plus Google Pay, tokenised saved cards and wallet/gift-card tender.
    • Supported methods wired end to end through checkout: KNET, Benefit, mada, STC Pay, Tami, Amex, Visa/Mastercard, Tabby BNPL, wallet balance, gift cards and cash on delivery.
    • 40+ Gradle feature modules — cart, checkout, order tracking, subscriptions with auto-renewal, wallet top-up, rewards, referrals, coupons and surge pricing.
    • Seven analytics and attribution SDKs isolated per module behind an EventDataParser seam so one vendor never leaks into feature code.
    • Apollo GraphQL per brand, REST over OkHttp, SQLDelight caching with migrations, Koin, Compose with MVVM/StateFlow.
    KotlinComposeMVIKoinApollo GraphQLSQLDelightPaymentsWhite-label

  2. Senior Android Developer

    Feb 2021 — Sep 2023

    Ozoned Digital · Lahore, Pakistan

    • Built server-driven UI that cut load times ~30% and let product ship without releases.
    • Integrated 5 payment gateways — three bank and two mobile-wallet providers behind one checkout — for +50% transaction success.
    • OCR for national ID / driving licence capture; Amazon Rekognition for AI vehicle damage assessment.
    • Health data sync across 4 health and wearable platforms.
    KotlinHiltRxJavaCoroutinesAI/MLPayments

  3. Senior Android Developer

    Sep 2017 — Feb 2021

    Xeon Technologies · Lahore, Pakistan

    • Improved sensor accuracy 25% with BLE tyre-pressure sensor communication.
    • Shipped OTA firmware upgrades driven straight from the Android app.
    • Firebase chat with image, video, audio, file and location payloads.
    • AdMob mediation & waterfall, Play Store policy compliance, Node.js APIs on the side.
    MVVMDaggerBLEOTAFirebase

  4. Android Developer

    May 2017 — Jul 2017

    RoboArt · Lahore, Pakistan

    • IoT device connectivity over Bluetooth, Google APIs, Firebase DB & notifications, location services.
    IoTBluetoothFirebase
03 — Fit check

Role requirement check

Select the capabilities a role requires, or paste the job description. The tool reports which requirements are fully covered, which are partially covered and which are not, with supporting evidence for each. It runs entirely in the browser; no data is transmitted.

Select the role requirements No requirements selected

Or paste the job description Optional
04 — Track record

Delivery record

A summary of delivery history across full-time roles and contract engagements.

Verifiable delivery metrics:

  • 100%job success across 56 contracts
  • 2,203hours tracked on Upwork
  • 36client reviews, 5.0 average
  • 3×contracts with a single returning client
  • 3 yrsfully remote in an Abu Dhabi Scrum team

Working practice: daily written status updates, committed estimates, releases gated on green CI, Crashlytics review after every rollout, Play policy compliance addressed before submission, and Spotless, Detekt, JUnit and MockK checks enforced on every branch.

05 — Clients

Top Rated Plus on Upwork

Senior Android contract work in Kotlin and Compose across white-label commerce, fintech and BLE, with a publicly verifiable record.

Top Rated PlusUpwork top-tier talent badge
0%Job Success Score
0jobs completed
0hours worked
0client reviews

Engagement types

  • Native Android
  • Kotlin · Compose
  • White-label apps
  • Payments & wallets
  • Health & wearables
  • BLE & OTA
  • XML to Compose migration
  • ANR & crash remediation

Working practice

  • Daily written updates
  • Committed estimates
  • Clean, testable code
  • JUnit · MockK
  • CI/CD releases
  • Play policy & crash monitoring

Client feedback

“As always, results as expected. Good communication, quick edits. Final result = 100%. Thank you!”

★★★★★ Repeat client · Compose migration Clear communicator

“Good communication. Code works as expected. All edits were made in a reasonable amount of time. This was our 3rd contract and I'm really satisfied. Thank you!”

★★★★★ Third contract · Android feature work Committed to quality

“Super communication. Job done really fast. Freelancer is very friendly, knows a lot about Android & Kotlin. I'm really satisfied! Definitely recommend. 10/10”

★★★★★ Wear OS app · smartwatch client Clear communicator

Verbatim client feedback from completed Upwork contracts, each rated 5.0 out of 5. Project titles are listed on the Upwork profile.

Upwork profile WhatsApp Email
06 — Apps

Shipped applications

All published on Google Play. Screenshots and install counts are taken directly from the public listings. Featured products are listed first, followed by the white-label portfolio and additional delivered work.

Barns app home screen showing the loyalty card and drink shortcuts Barns app menu screen with hot and cold drink categories and prices in SAR Barns app order screen
2023 → now · COFE App, Abu Dhabi
Barns app icon

Barns

Food & drink · Saudi Arabia

100K+ installs from a single white-label build

Android application for one of Saudi Arabia's largest coffee chains, covering menu browsing, drink customisation, pickup and delivery, loyalty and payments. Built on a white-label architecture in which a configuration change produces a fully branded application.

100K+downloads View on Google Play
How it was built

The constraint. A branded app per coffee chain, delivered on the schedule of a marketing team rather than an engineering one. Forking the repository per brand would have meant every bug fix becoming a merge across every fork.

The decision. Branding was moved out of feature code entirely and treated as configuration: palette, endpoints and an enabled feature set resolved once at the composition root. Feature code asks what capability it needs, never which brand it is running as.

The result. Adding a brand became a configuration change rather than a code change, and one release train serves the whole portfolio. Barns alone has passed 100K installs on this build.

KotlinComposeClean ArchitectureMVIKoinRetrofit
COFE app home screen listing nearby coffee brands COFE app brand menu with drink categories COFE app order and checkout screen
2023 → now · COFE App, Abu Dhabi
COFE app icon

COFE

Food & drink · Kuwait, KSA, UAE, Turkey

500K+ installs · 2 gateway SDKs, 12+ brand apps, one codebase

The flagship marketplace from which the white-label platform originated: dozens of coffee brands in a single application, with delivery, click-and-collect, virtual drive-thru, catering, in-app credit and cross-brand rewards across Kuwait, Saudi Arabia, the UAE and Turkey.

500K+downloads View on Google Play
How it was built

The constraint. Checkout ran across multiple payment gateway SDKs, each reporting failure in its own vocabulary — exceptions from one, status strings from another, some reasons arriving only by webhook. No shared retry policy could be written against that.

The decision. Every gateway was placed behind one sealed result type that distinguishes a genuine decline from a recoverable, transient failure. Transient failures are retried; only real declines are shown to the customer.

The result. Customers were previously told a payment had failed when the network had simply hiccupped, and most did not try a second time. Correcting that lifted transaction success by roughly 50%.

KotlinComposeMarketplacePaymentsLoyalty
Jubilee Active dashboard showing daily steps and reward points Jubilee Active rewards screen Jubilee Active activity tracking screen
2022 → 2023 · Ozoned Digital, Lahore
Jubilee Active app icon

Jubilee Active

Health & fitness · Pakistan

4 health platforms synced — Google Fit, Health Connect, Samsung Health, Garmin

Health and fitness tracking application for Jubilee Life Insurance. Step and calorie tracking integrated with Google Fit, Health Connect, Samsung Health and Garmin, with walking challenges and activity points redeemable against policy discounts.

100K+downloads View on Google Play
How it was built

The constraint. Activity points converted into real policy discounts, so the step count had to be defensible. Reading Google Fit, Health Connect, Samsung Health and Garmin and summing them counted the same walk two or three times over.

The decision. Each platform sits behind one ActivitySource interface with a trust priority, and every reading is normalised by a dedicated mapper. Overlapping time windows are resolved once in the domain layer by that priority, so a walk recorded by both a watch and a phone becomes a single entry.

The result. One reconciled figure per day that both the insurer and the customer could rely on, with new platforms added by implementing a single interface.

Health ConnectGoogle FitGarminSamsung Health
Illustration of AI vehicle damage detection
AWS & Epiphany · 2nd of 25

AI vehicle damage assessment

Computer vision · competition build

+80% recognition accuracy · 2nd of 25 companies nationally

Camera-based vehicle damage assessment built on Amazon Rekognition with on-device capture and OCR pipelines. Delivered 80% higher recognition accuracy and placed 2nd of 25 companies nationally in the AWS and Epiphany AI/ML Startup Reactor competition.

Competition project · not publicly released
Amazon RekognitionAI/MLOCRCameraX
Payments

Payment integrations delivered

Two production gateway SDKs sit behind a single PaymentProvider abstraction, so entering a new market requires adding a provider implementation rather than modifying checkout. Every item below is wired end to end through a live checkout.

Gateway SDKs integrated

  • Tap Payments GoSell
  • Checkout.com Frames
  • Google Pay
  • Google Play Billing

Tokenisation, charge and authorise, saved cards, themed SDK sessions, environment switching, network error handling.

Supported payment methods

  • KNET
  • Benefit
  • mada
  • STC Pay
  • Tami
  • Amex
  • Visa · Mastercard
  • Tabby BNPL
  • Apple Pay scheme
  • Wallet balance
  • Gift cards
  • Cash on delivery

Includes add-card and tokenised saved-card flows. Tabby is routed through the gateway and the Apple Pay scheme exists for backend parity; neither is an Android SDK integration on this side.

White-label platform

Brand applications on the shared platform

One Kotlin/Compose codebase carries 12+ brand modules, each configured by its own vendor configurator into a separate build with distinct identity, menu, loyalty rules and payment setup. The following brands were confirmed live on Google Play:

Unlisted brand modules

Additional brand modules on the platform

These brands also have modules in the same codebase, but no live Google Play listing published by COFE could be confirmed. They are recorded here as engineering work rather than linked as published applications.

French BakeryPapparotiSubwaySurge
Earlier and adjacent work

Other delivered applications

Additional delivered work across insurtech, retail and connected hardware, all currently or previously published on Google Play.

Download figures above are Google Play's published install ranges, recorded from the live listings on 31 Aug 2026. They describe each application as a whole, including store operations, backend, content and marketing, rather than the contribution of any single engineer.

07 — Toolkit

Technical stack

Android-first, with Kotlin Multiplatform and Compose Multiplatform in production: one Compose codebase published on Google Play across Android, iOS and Wasm web. Every item below has been used in production, and each links to its official documentation for verification.

Cross-platform

Architecture

  • Clean Architecture
  • SOLID
  • Dependency inversion
  • MVVM · MVI
  • Multi-module (50+ modules)
  • Module boundaries
  • Testability by design
  • Convention plugins
  • Server-driven UI

Analytics & attribution

Engineering practice

Clean Architecture and SOLID in the delivered code

Architecture outlasts any individual feature. The following describes how these principles appear in the delivered codebases; each statement corresponds to work documented elsewhere on this page.

Clean Architecture in practice

Domain logic resides in pure Kotlin modules with no framework imports, making it testable without an emulator and reusable across Android, iOS and web targets. Data sources sit behind repository interfaces owned by the domain layer; UI holds state and never business rules. Feature modules depend on abstractions rather than on each other, which is what allows 12+ brand applications and 40+ feature modules to be delivered from a single codebase without a merge queue.

SOLID in application

Single responsibility: mapping resides in dedicated mappers rather than in enums, view models or UI. Open/closed: entering a market requires adding a PaymentProvider implementation, not editing checkout. Interface segregation: seven analytics SDKs sit behind one narrow EventDataParser seam, so no vendor type reaches feature code. Dependency inversion: platform services are injected through expect/actual or dependency injection, never referenced directly.

Adoption of new platforms

New technologies are adopted through production delivery. Kotlin Multiplatform progressed from unfamiliar to a published Play release — one Compose codebase on Android, iOS and Wasm web with a Postgres rules backend — under a project deadline rather than through training. The same approach applied to Compose, Health Connect, BLE and OTA, and Play Billing: work from the primary documentation, build the smallest production-grade implementation, then harden it.

AI scope

Machine learning in production and AI-assisted engineering

Two distinct areas, stated separately so the scope of each is unambiguous.

Computer vision in production

An Amazon Rekognition pipeline assessing vehicle damage from photographs, which raised recognition accuracy by 80%; the team led placed 2nd of 25 companies nationally in the AWS and Epiphany AI/ML Startup Reactor. Also includes OCR pipelines for CNIC and driving-licence capture, and the face recognition module in Classmate Playdate.

AI-assisted engineering

Daily development with AI coding agents, supported by codebase structure that makes agent work reviewable: strict module boundaries, convention plugins, deterministic builds and quality gates that fail explicitly.

Scope of experience

No production LLM, RAG or agent product to date. The activity planner in Classmate Playdate is a rules-based feature with no external model integrated. For LLM-focused roles, the relevant experience is mobile engineering depth and daily use of these tools rather than a shipped generative AI product.

08 — Credentials

Education, certifications & recognition

Recognition

Certificates

Signed awards, viewable without a login. Open any image for the full-size version.

09 — Contact

Contact

Open to senior mobile engineering roles, remote or with relocation to Europe and the GCC, and to selected contract engagements. Enquiries are answered the same day, typically within a few hours.

Book a 30-min call
WhatsApp +92 300 4534531
Email ahsanislam04@gmail.com
LinkedIn GitHub Upwork Download résumé Résumé (Markdown)