Flutter, native apps
and connected devices.

A mobile vision-training application that connects BLE training hardware with interactive web content. The engineering work connects device events, application state and training sessions.

Dart, Flutter, GetX, FlutterBluePlus, WebView

• Device integration

  Worked on BLE scanning, connections, GATT notifications and device commands for lens control, training settings and battery status. Buffered incoming data to assemble offline training records for upload.

• Connection behavior

  Implemented remembered-device reconnection and connection-state updates. The workflow respects a user's manual disconnect choice instead of immediately reconnecting during the same app session.

• Training workflow

  Integrated a JavaScript bridge for training configuration, lens-flip commands and device callbacks. Connected training completion to result submission and handled network errors in the application.

Delivered application features linking training content, connected hardware and training records.

Enabled existing native Android applications to use Flutter training screens and BLE features through a native API.

Flutter Module, Kotlin, FlutterEngine, MethodChannel, Android Library / AAR

• Runtime and API design

  Initialized and cached a FlutterEngine for reuse when opening training, device and report screens. Kept the engine alive when an SDK screen closed. Used MethodChannel for routing, tokens, parameters, BLE commands and device-state callbacks.

• Packaging and integration

  Delivered an Android library AAR together with the Flutter module's generated Maven repository and dependencies. Provided native methods and callbacks so host applications could integrate the features without implementing the internal Flutter workflows.

Delivered a reusable native integration package for Android host applications.

Independently built and delivered the Android application for a vision assessment and eye-tracking device, covering architecture, core workflows, hardware integration and troubleshooting. The product has been deployed on several thousand devices.

Kotlin, Jetpack Compose, vendor eye-tracking SDK, BLE, WebView

• Application architecture

  Built Compose screens and ViewModels for assessment, training and report workflows. Coordinated application state with connected hardware and embedded web content.

• Eye-tracking integration

  Integrated a vendor SDK for device initialization, calibration and gaze-data callbacks. Connected the SDK output to assessment and training behavior; the underlying eye-tracking algorithm was supplied by the vendor.

• BLE device control

  Implemented native BLE discovery, GATT connections, notifications and device commands. Coordinated connection-state changes and a heartbeat mechanism with the training workflow.

• Web and native communication

  Built a JavaScript action bridge between embedded training content and Android capabilities, connecting web events to native device operations and application state.

• Report export and sharing

  Rendered Compose report views off screen to create report images, generated A4 PDFs and supported Android sharing. Added an image-upload and QR-code workflow for downloading reports outside the device.

• Delivery and diagnostics

  Handled requirements analysis, architecture, core implementation, logging, exception monitoring, integration testing, on-site troubleshooting and final application delivery.

Delivered the Android application from requirements through device integration and deployment support.

An AI background-removal and image-editing application. My work focused on the native iOS editor and its GPU rendering path, addressing multi-layer editing lag and brush-precision limitations when editing 4K images.

iOS, Core Image, Metal, render pipelines, compute / fragment shaders

• Renderer migration

  Led the migration of the core editor from Core Image to Metal. Built GPU rendering with textures, command buffers, render pipelines and compute / fragment shaders.

• Editing tools

  Implemented precise brushes and mask erase / restore for pixel-level image editing. Integrated these interactions with the editor's rendering workflow.

• Undo and redo

  Implemented editing history using saved states and a current-state pointer, allowing users to move backward and forward through editing operations.

Delivered the Metal-based core editor and editing tools for 4K image workflows.

A video-template application with dozens of screens. I developed core Flutter and iOS features, contributed to editor architecture and supported the gradual migration of native modules to Flutter.

Four-person development team | Dart, Flutter, native iOS, Platform Channels

• Incremental migration

  Contributed to module-by-module migration and Flutter-native communication, enabling Flutter features to work with existing native capabilities.

• Shared capabilities

  Independently implemented cross-platform login and Alibaba Cloud analytics integration.

Delivered Flutter and native iOS features while contributing to the product's incremental migration.