Mobile Developer for Real-Time 3D Talking Avatar App

We are looking for an experienced **mobile developer with strong real-time 3D rendering experience** to help build an Android application featuring a live animated 3D avatar. This is **not a 3D artist, character animator, or 3D modeling position**. We expect to provide the 3D avatar assets, including the GLB or other supported 3D model format, rig, textures, materials, and ARKit-compatible facial blendshapes. The main responsibility is to develop the mobile application, integrate face tracking, render the 3D avatar in real time, and ensure excellent visual quality and smooth performance on Android devices. ## Project Overview The application will use the phone’s front-facing camera to track the user’s face and animate a personalized cartoon-style 3D avatar. The avatar should follow the user’s: * Head rotation and position * Eye movement * Blinking * Eyebrow movement * Mouth movement * Facial expressions * Selected upper-body movement, when supported The avatar will use approximately 52 ARKit-compatible facial blendshapes. For example: * When the user smiles, the avatar should smile. * When the user blinks, the avatar should blink. * When the user opens their mouth or talks, the avatar’s mouth should move naturally. * When the user turns or tilts their head, the avatar should follow smoothly. * When the user changes facial expressions, the avatar should respond in real time. The app should display the animated avatar instead of, or together with, the raw camera preview. ## Main Responsibilities * Develop the Android application. * Integrate the front-facing camera. * Integrate a real-time face-tracking solution. * Convert facial-tracking results into ARKit-compatible blendshape values. * Apply blendshape values to the 3D avatar in real time. * Apply head and neck rotation to the avatar. * Implement natural eye movement and blinking. * Support real-time mouth and facial animation while the user is talking. * Render GLB, glTF, VRM, or another suitable 3D avatar format. * Create smooth transitions between facial expressions. * Reduce jitter and unstable movement from face-tracking data. * Optimize the application for stable Android performance. * Implement high-quality lighting, shadows, materials, and camera presentation. * Profile CPU, GPU, memory, frame time, and rendering performance. * Test and optimize the application on real Android devices. * Build a maintainable architecture that can support additional avatars and customization options. ## Most Important Requirements ### 1. Mobile Application Development You should have strong experience developing production Android applications using one or more of the following: * Kotlin * Java * Unity * Unreal Engine * Flutter with native rendering integration * React Native with native rendering integration * Another suitable mobile development framework Native Android, Unity, or another solution capable of high-performance real-time 3D rendering is preferred. ### 2. Real-Time 3D Rendering on Android You should have experience integrating 3D rendering into a mobile application using technologies such as: * Google Filament * OpenGL ES * Vulkan * Unity * Unreal Engine * SceneView * Custom rendering engines * Other production-ready mobile 3D frameworks You should understand: * Mobile GPU limitations * Rendering pipelines * Physically based materials * Shaders * Lighting * Shadows * Anti-aliasing * Texture optimization * Draw-call optimization * Frame pacing * Device-specific GPU behavior ### 3. Live Talking Avatar or Facial Animation Experience with real-time talking avatars, virtual characters, digital humans, VTubers, AR characters, or facial-animation systems is highly preferred. Relevant experience may include: * ARKit 52 facial blendshapes * MediaPipe Face Landmarker * ARCore * Face landmarks * Facial-expression tracking * Morph-target animation * Blendshape animation * Lip movement * Eye tracking * Head-pose tracking * Real-time character animation You should understand how to map live camera-tracking data to a rigged 3D avatar. ### 4. Motion Smoothness The facial and head movement must look smooth and natural. The implementation should minimize: * Facial jitter * Head shaking * Sudden movement * Blendshape popping * Delayed reactions * Unnatural blinking * Unstable mouth movement * Frame-rate-dependent animation Experience with interpolation, filtering, quaternion smoothing, blendshape smoothing, animation blending, and latency reduction is important. ### 5. Rendering Quality Visual quality is a major priority. We want the avatar to have a polished, premium cartoon appearance rather than looking like a simple mobile-game character. The developer should be able to improve: * Lighting * Soft shadows * Skin appearance * Hair appearance * Eye reflections * Teeth and mouth rendering * Material quality * Tone mapping * Anti-aliasing * Environment lighting * Overall visual presentation The visual target is a high-quality animated-movie-style cartoon character while still maintaining practical mobile performance. The final character design must be original. We are using “animated-movie quality” only as a reference for polish, softness, lighting, and expressiveness. ## Avatar Assets We expect to provide a prepared avatar asset, potentially including: * GLB, glTF, or VRM model * Skeleton * Head and neck bones * Textures * Materials * ARKit-compatible blendshapes * Hair * Eyes * Teeth * Clothing * Accessories You are not expected to create the character from scratch. However, experience inspecting GLB, glTF, VRM, or Blender assets is preferred because you may need to identify technical issues affecting rendering or animation. For example: * Incorrect blendshape names * Unsupported materials * Transparency problems * Excessive polygon count * Large textures * Too many draw calls * Incorrect bone hierarchy * Poor mobile performance * GLB compatibility problems ## Avatar Customization The application should eventually support multiple avatar customization options, including: * Hairstyles * Hair colors * Face structures * Eyes * Eyelids * Eyebrows * Noses * Mouths * Skin tones * Facial hair * Clothing * Glasses * Other accessories The developer should design the application so these options can be switched dynamically. Experience building modular avatar or character-customization systems is preferred. The customization system must remain compatible with: * Facial blendshapes * Head movement * Real-time animation * Lighting * Materials * Android performance requirements ## Future Avatar Generation Feature In a later phase, we would also like to explore generating or configuring a personalized avatar from: * A user photo * Multiple selfies * A short selfie video This may involve identifying the user’s: * Face shape * Skin tone * Hairstyle * Eye shape * Eyelids * Nose * Mouth * Other visible facial characteristics The first version does not necessarily require full automatic 3D avatar generation. However, experience with photo-to-avatar systems, face analysis, 3D face reconstruction, computer vision, or avatar-generation APIs would be valuable. ## Performance Expectations The application should target: * Stable real-time rendering * Approximately 30 FPS or higher on supported Android devices * Consistent frame pacing * Low facial-tracking latency * Smooth animation * Reasonable memory usage * Reasonable battery consumption * No frequent freezing or frame drops The developer should be able to profile: * CPU frame time * GPU frame time * Memory usage * Texture memory * Draw calls * Shader cost * Blendshape update cost * Bone-animation cost * Camera and tracking latency * Motion-to-display latency * Thermal throttling ## Preferred Skills * Android development * Kotlin or Java * Unity or Unreal mobile development * Google Filament * OpenGL ES or Vulkan * MediaPipe Face Landmarker * ARKit-compatible blendshapes * ARCore * GLB, glTF, or VRM * Morph targets * Real-time facial animation * Talking-avatar applications * Mobile GPU optimization * Custom shaders * Character-customization systems * Three.js or WebGL experience as an additional skill Blender experience is helpful, but this is not a Blender artist or modeling role. ## Initial Paid Prototype The first milestone will be a small paid prototype. We will provide a sample 3D avatar with facial blendshapes. The prototype should demonstrate: * Loading the avatar in an Android application * Rendering the avatar with high-quality lighting * Using the front camera for face tracking * Mapping facial tracking to ARKit-compatible blendshapes * Head and neck movement * Blinking and eye movement * Mouth movement while talking * Basic facial expressions * Smooth motion with minimal jitter * Performance testing on a real Android device The developer should also provide: * FPS results * CPU and GPU frame times * Memory usage * Tested device information * Identified technical limitations * Recommended rendering architecture * Recommendations for the next production phase ## When Applying Please include: 1. Examples of mobile applications where you personally implemented real-time 3D rendering. 2. Examples of talking avatars, digital humans, facial animation, AR characters, or VTuber-style applications. 3. The mobile rendering technologies you used. 4. Your experience with ARKit blendshapes or facial morph targets. 5. Your experience with MediaPipe, ARCore, or similar face-tracking technologies. 6. How you would keep head and facial movement smooth. 7. How you would achieve high-quality cartoon rendering on Android. 8. Performance measurements from a previous mobile 3D project. 9. The Android devices and GPU types you have tested. 10. Which rendering technology you recommend for this application and why. Please begin your proposal with the words **“Live 3D Avatar”** so we know you reviewed the full description. We are looking for a **mobile engineer with real-time 3D graphics experience**, not a 3D artist or offline character animator.