Skip to content
View NazarKozak's full-sized avatar
🎯
Focusing
🎯
Focusing
14 followers · 13 following

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx3Achievement: YOLOAchievement: StarstruckAchievement: Arctic Code Vault Contributor

Achievements

Achievement: Pair Extraordinairex3Achievement: QuickdrawAchievement: Pull Sharkx3Achievement: YOLOAchievement: StarstruckAchievement: Arctic Code Vault Contributor

Block or report NazarKozak

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
NazarKozak/README.md

Hi, I'm Nazar

Software Engineer · IEEE Senior Member · Los Angeles, CA

I build mobile AI systems — from on-device ML inference to SDK architectures serving millions of users.

What I'm working on

  • Disfluo — On-device speech disfluency detection & voice stress analysis framework for iOS. CNN models running in <1ms via CoreML. Paper under review at IEEE Signal Processing Letters.
  • Secret Project — GenAI-powered agentic features for the App.

Selected work

Project Role Impact
Solitaire Co-Creator & Lead Dev 10M+ downloads, 200K DAU, $8M+ revenue
Appodeal SDK Founding Engineer Unity SDK powering 80% of company revenue; company grew to 300+ employees
SeaGreen (Datasya) Head of Mobile 10x inference improvement, real-time respiratory monitoring, contributed to $3M seed
TFLiteMovement Creator Open-source TensorFlow Lite pose detection on iOS/SwiftUI

Research

  • N. Kozak, "On-Device Disfluency Detection with Voice Stress Association Analysis: A Mobile Framework Validated on SEP-28K," submitted to IEEE Signal Processing Letters, 2026.
  • N. Kozak, "Dynamic Multidex Loading for Cross-Platform SDK Integration," Zenodo, 2026. DOI: 10.5281/zenodo.19197878

Tech

Swift SwiftUI CoreML PyTorch Python Accelerate/vDSP ARKit Metal React Native Flutter Unity

ORCID · IEEE · Toptal · nzrkzk@gmail.com

Pinned Loading

  1. TFLiteMovement TFLiteMovement Public

    Real-time human pose detection on iOS using TensorFlow Lite and SwiftUI

    Swift 8

  2. cordova-plugin-multidex cordova-plugin-multidex Public

    Dynamic multidex loading plugin for Cordova — enables runtime DEX loading on Android

    JavaScript 3 5

  3. SurfaceRecorderSDK SurfaceRecorderSDK Public

    Record anything on iOS to video — an AR scene, the camera, or any view — with synced audio, without ReplayKit's permission prompt. Swift 6, GPU-fast.

    Swift

  4. RepCounterSDK RepCounterSDK Public

    Count exercise reps on-device, for free — no SDK key, no cloud. Apple Vision body-pose rep counting with hysteresis + dual-joint anti-cheat. The open alternative to paid pose-fitness SDKs.

    Swift

  5. OnDeviceVLMSDK OnDeviceVLMSDK Public

    On-device image understanding in a few lines of Swift — caption, VQA, tags, OCR with a small VLM (SmolVLM/FastVLM/Qwen2.5-VL) via MLX. No cloud, no API keys. WhisperKit, but for vision.

    Swift

  6. MLXPose MLXPose Public

    Native MLX Swift ViTPose: on-device human pose estimation on Apple Silicon (HF-verified). No PyTorch, no cloud.

    Swift