Skip to content

Latest commit

 

History

68 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 

Repository files navigation

Lviv, Ukraine 🇺🇦
LinkedIn

Telegram

github.com/dabakovich

dabakovich@gmail.com

About Me

Senior React Native Engineer with 6+ years across the full mobile stack — React Native and TypeScript up the stack, Swift/Kotlin and TurboModules down it. Tech-lead and people-manager experience (3–6 direct reports) on top, plus a public open-source footprint outside work. Agent-native for 1.5+ years: ~98% of my code is now agent-generated, driven through disciplined planning, custom skills, and MCP — not vibe-coding, but engineered agent workflows.

Selected Achievements

  • Built and maintained Guesty mobile apps used daily by thousands of property managers worldwide across iOS + Android.
  • Cut Guesty app initial load 78% (700 ms → 150 ms) and tripled scroll FPS by building a virtualized 2D grid for React Native.
  • Shipped 50+ store releases with zero critical rollbacks across iOS + Android.
  • Author of react-native-controlled-mentions — 30k+ weekly npm downloads, used in production by multiple companies.
  • Volunteer contributor to Smart Event Detector (2024) — Ukraine-defense acoustic-triangulation system that locates battlefield explosion sites via sound.
  • Pioneered reliable autonomous agent loops for React Native — wired Claude Code to Maestro e2e tests so the agent runs the app on-device, reads real UI results, and self-corrects without a human in the loop.
  • Built my own coding agent (ai-project-assistant) in early 2024, before Cursor/Claude Code were mainstream; now run ~98% agent-generated code across production work with 1.5+ years of regular agentic development.
  • Mentored 3–6 engineers with weekly 1:1s, PDPs.

Agentic AI Development

Two years running production development through AI agents, evolving from a custom-built agent to today's tooling. The workflow is engineered, not ad-hoc:

  • Planning-first: decompose work into specs and plans before the agent writes code.
  • Skills: compose reusable agent skills — primarily superpowers-based, with other tech-specific or custom skills per project.
  • MCP: integrate tools and context sources via Model Context Protocol.
  • Agent loops: closed-loop self-correction, including a Maestro-based approach for real React Native e2e verification on device.
  • Tooling: Claude Code, Codex, and Cursor — matched to the task.

Skills And Technology

Mobile (Primary): React Native, TypeScript, react-navigation, Reanimated, Redux Toolkit, MobX, MMKV/AsyncStorage, Jest, Detox, Firebase, App Store Connect, Google Play Console

Native bridging: Swift, Kotlin, Objective-C basics, TurboModules, custom native modules (including New Architecture)

Mobile delivery: CI/CD (Fastlane, GitHub Actions), OTA updates (CodePush), crash reporting, A/B testing, feature flags, Push Notifications, Deep Linking

Agentic AI development: Claude Code, Codex, Cursor · agent skills (superpowers + custom) · MCP · spec/plan-driven workflows · closed-loop agents (Maestro e2e for RN) · 2+ years, ~98% agent-generated code

Frontend: React, Redux Toolkit, MobX, Material UI, React Router

Backend: Node.js, Express, NestJS, REST, MongoDB/Mongoose

Practices: Performance profiling (Flipper, RN DevTools), TDD, code review, performance profiling, i18n

Other interests

ML/AI: Python, NumPy, Pandas, scikit-learn, TensorFlow, Reinforcement Learning, genetic algorithm LLM-based agents, LangChain, LangGraph, RAG

Leadership

  • Managed 3–6 engineers across two squads; ran weekly 1:1s, PDPs, and quarterly reviews.

  • Hired 2 engineers end-to-end (sourcing → tech screen → offer);

  • Tech-lead duties: architecture decisions, best practices implementation and control, mentorship.

Work Experience

Senior React Native Engineer Guesty Mobile Apps at Forbytes

May 2022 - May 2026 · iOS + Android · Hospitality / PMS platform

Hospitality Management Platform

As a Senior React Native Engineer at Guesty, I played a critical role in developing and maintaining the "Guesty" and "Guesty for Hosts" mobile apps. These apps enable property managers to manage reservations, automate guest communication, and streamline operations. Some of my key contributions to this project included:

  • Managed migration legacy code to application from scratch, decreased initial load from 2s to 600ms.
  • Shipped 50+ App Store / Play Store releases with zero critical rollback.
  • Owned key flows: core Multi-Calendar feature and Tasks/Listings features.
  • Implemented virtualized grid (like FlatList but 2d), optimized initial load from 700ms to 150ms, increased scroll FPS from 20 to 60.
  • Collaborating with the team to develop and implement new features, such as integrating third-party APIs for payment processing and booking management.
  • Developing and maintaining a robust codebase using best practices and standards, with a strong emphasis on code quality and testing.
  • Partnered with design + product on a shared component system.

Mobile / Full-Stack Engineer (later People Manager) — Rake at 42flows.tech (previously Chatbots.Studio)

September 2018 - May 2022

Allows businesses to connect external platforms (Twilio, FB, WhatsApp, web widgets) to the workspace in Rake and communicate with their customers.

  • Worked with the team on both backend and frontend side.
  • Completely developed mobile application using React Native and TypeScript.
  • Owned CI/CD on App Center → migrated to Fastlane, cutting release time from 2h to 25m.
  • Drove TypeScript strict-mode rollout across the codebase, eliminating any from 400+ files.
  • Mentored 3 mid-level engineers; ran weekly tech-review calls and PDP cycles.

Projects

DevYogaPet Project — AI task tracker built in one week · Jun 2026

Full-stack task tracker for developers with a built-in conversational AI agent that manages the board through chat. React Native app on the latest Expo SDK 56 with native SwiftUI components, NestJS backend — designed, built, and shipped in one week.

  • Built a multi-step AI agent (Claude via Vercel AI SDK) with tool-calling for task creation, "plan my day" prioritization, and triage — with human-in-the-loop confirmation before writes and persistent agent memory.
  • Implemented stateless chat: the transcript is persisted on-device (MMKV) and replayed per request, keeping the backend session-free.

Smart Event DetectorVolunteer project — Ukraine defense · Feb 2024 – Jul 2024

Distributed acoustic-triangulation system that pinpoints battlefield explosion sites by measuring the arrival-time delta of the shockwave across a network of Android devices — conceptually GPS, but with sound waves instead of radio. The React Native app continuously analyzes the raw audio stream, detects explosion events locally with high-precision timestamps, and forwards them to the backend, which solves for the source coordinates via time-difference-of-arrival (TDoA).

  • Built the React Native client: continuous raw-audio capture, on-device explosion detection, and precise event timestamping for server-side multilateration.
  • Authored a custom TurboModule (New Architecture) to run the real-time sound-event detection off the JS thread, cutting per-batch processing from ~60 ms to <1 ms (60× faster) — essential for the sub-millisecond timestamp accuracy the TDoA math requires.
  • Designed the mobile↔backend event protocol so multiple devices can stream synchronized detection events for coordinate triangulation.
  • Integrated Firebase Realtime Database for low-latency device presence and live GPS tracking, giving operators an instant map view of which devices are online and where on the battlefield.

react-native-controlled-mentionsReact Native community library 30k+ NPM weekly downloads, ★ 250+ on GitHub

Widely-used React Native library for mentions and rich text inside TextInput, no native modules required. 30k+ weekly npm downloads, 250+ GitHub stars, used in production by multiple companies. Originally built for Rake, then open-sourced.

ai-project-assistantCustom AI coding agent — Jan 2024

Autonomous coding assistant I built to navigate, read, and modify a codebase and run Git operations from natural-language tasks — a self-made agent that predated mainstream tools like Cursor and Claude Code, and shaped how I work with agents today.

  • Implemented an agent loop on OpenAI GPT models with tool/function-calling to drive file navigation, read/create/modify operations, and Git diffs.
  • Architected around clear modules — AI interaction, file-system operations, and shared utilities for message generation and tool-call handling.
  • Built entirely in TypeScript on Node.js, with type-safe tool definitions bridging the model to the local project.

Neural Network from ScratchPet Project

Custom neural network implementation built from scratch using NumPy with full back-propagation algorithm. The project includes a reinforcement learning experiment where two AI agents learn to play TicTacToe through self-play.

  • Implemented complete neural network with support for multiple activation functions (ReLU, Leaky ReLU, Sigmoid, Tanh, Softmax) and proper weight initialization strategies (He, Xavier).
  • Developed back-propagation algorithm with support for both MSE and Log loss functions, including special Jacobian matrix handling for Softmax layers.
  • Built reinforcement learning system with two competing agents (9→18→18→9 architecture) that learn optimal TicTacToe strategies through reward-based training over thousands of self-play games.
  • Implemented real-time statistics tracking and visualization to monitor agent performance, win rates, and learning progression.

Earlier Projects

Strobe SpectroscopyController

Early student project. Application with Swing GUI, that control motor stepper through Arduino, and reading analog data from the micro-controller. This allows to computerize instrument for measuring light spectrum.

Education

Ready Tensor — Agentic AI Developer Certification Program

May 2025 - July 2025

  • Built agentic systems using LangChain with RAG (Retrieval-Augmented Generation) pipelines and vector database integration.

robot_dreams – How to build a recommendation model in Python in 2 days

August 2024

Stratoplan Manager School — Course "Team"

September 2021 - February 2022

  • Team audit on entry
  • Communication within the team (constructive confrontations)
  • Interviewing, hiring and firing
  • Setting and controlling tasks, delegation
  • Team status monitoring and feedback

Logos, Lviv — Java (Hibernate, JPA, Spring)

April 2016 - September 2016

LNU, Lviv — LNU, Master’s degree, Electronics

September 2011 - July 2017

Languages

  • Ukrainian: Native
  • English: B2, comfortable with daily standups, written communication, and code reviews

About

My personal curriculum vitae

Resources

Stars

0 stars

Watchers

1 watching

Forks

Releases

Packages

Contributors