MAXIM BUNKOV

AI-Native Product Engineer

Remote · Europe

Senior software engineer with 12+ years of experience building production software across Apple platforms, mobile, developer tooling and systems engineering.

I build products end-to-end — from an ambiguous idea and requirements to specification, architecture, implementation, testing and production.

My core expertise is Swift, iOS/macOS and React Native, complemented by Rust, TypeScript, Python, networking, real-time media and developer infrastructure.

I work with a specification-driven, agentic development workflow built around Kiro Spec-Driven Development, Oh My Pi and Zed.

I also build my own AI development infrastructure, including custom MCP servers and a private semantic knowledge system backed by PostgreSQL and pgvector.

AI-NATIVE ENGINEERING

Kiro · Oh My Pi · Zed · MCP · PostgreSQL · pgvector

I treat AI agents as active engineering collaborators rather than code autocomplete.

My development workflow combines structured specifications, agent orchestration, custom tools and a persistent semantic knowledge layer.

Development workflow

Idea → Requirements → Kiro Specification → Architecture → Agentic Implementation → MCP Tools / Knowledge → Validation → Review → Production

Kiro Spec-Driven Development

I use Kiro's specification-driven approach to turn ambiguous product requirements into structured engineering specifications, acceptance criteria and implementation plans before implementation begins.

This gives both human developers and AI agents an explicit contract to work against.

Oh My Pi

I use Oh My Pi as the primary agentic development environment for orchestrating AI-assisted engineering tasks.

I use multiple agents and isolated Git worktrees when tasks can be developed independently, allowing parallel exploration and implementation.

Custom MCP Infrastructure

I build custom MCP servers to expose project-specific capabilities and context to AI agents.

Instead of relying exclusively on generic tools, I create MCP integrations around the actual development workflow and the information an agent needs to perform useful engineering work.

Semantic Knowledge Base

I maintain a personal semantic knowledge system backed by:

PostgreSQL + pgvector

The database stores project knowledge and embeddings, allowing relevant context to be retrieved semantically rather than relying only on keyword search.

This provides persistent project context that can be accessed by AI tooling and custom MCP integrations.

Zed

Zed is my primary development environment.

I use it together with my agentic workflow for direct code editing, exploration and verification.

CORE ENGINEERING

Languages

Swift · Rust · TypeScript · JavaScript · Python · Objective-C · Objective-C++

Apple Platforms

iOS · macOS · SwiftUI · UIKit · AVFoundation · AVPlayer · ScreenCaptureKit · ARKit · RealityKit

Mobile

React Native · Native Modules · CocoaPods · Swift Package Manager

AI & Developer Infrastructure

MCP · AI Agents · Agentic Workflows · Kiro Spec-Driven Development · Oh My Pi · pgvector · PostgreSQL · Semantic Search · LLM Applications

Systems

Networking · Real-time Communication · Streaming · Media Pipelines · Background Services · CLI Tools

Infrastructure

Git · GitHub · CI/CD · Bash · Automation · Developer Tooling

SELECTED PROJECTS

AI Development Infrastructure

MCP · PostgreSQL · pgvector · AI Agents · Semantic Search

Building a custom AI development environment around my own engineering workflow.

The system combines:

  • Custom MCP servers
  • Persistent project knowledge
  • PostgreSQL
  • pgvector embeddings
  • Semantic retrieval
  • AI agents
  • Kiro specifications
  • Git-based development
  • Automated engineering workflows

The goal is to give agents access to relevant project knowledge and tools while keeping the development process structured and reproducible.

Rather than repeatedly providing context to an AI model, relevant information can be retrieved from a persistent semantic knowledge base through custom tooling.

WireDeskVR

Swift · macOS · ScreenCaptureKit · Networking · VR

Experimental virtual workspace exploring low-latency desktop streaming into VR.

Designed and implemented a macOS background service responsible for screen capture, video processing and network communication with external devices.

The project combines macOS system programming, Swift, media processing, networking and VR.

KnowledgeQuery

AI · Developer Tooling · Git · Specifications · Semantic Knowledge

Experimental developer tooling focused on connecting project knowledge, specifications and implementation.

The project explores how structured specifications and persistent project knowledge can become useful context for AI-assisted engineering.

The underlying concept is:

Requirements → Knowledge → Context → Implementation → Verification

syncLproj

Rust · CLI · Apple Localization · CI/CD

Zero-dependency Rust command-line utility for synchronizing Apple .strings localization files.

Designed for deterministic CI/CD execution while preserving translations, ordering and developer comments.

SpeedReaderRust

Rust · CLI · Productivity

RSVP-based reading tool designed to accelerate processing of technical documentation and engineering specifications.

Built around a practical developer workflow problem: reducing the time and cognitive overhead required to consume large amounts of technical information.

ENGINEERING PHILOSOPHY

Specifications before implementation

I prefer making requirements explicit before writing large amounts of code.

Specifications provide a shared contract between product requirements, developers and AI agents.

Persistent context

AI systems become significantly more useful when they have access to relevant project knowledge.

I therefore prefer persistent, searchable knowledge over repeatedly rebuilding context inside individual conversations.

Build the tools you need

When an existing tool does not fit the workflow, I build the missing layer myself — whether that means a CLI, MCP server, automation or database-backed service.

Agents are collaborators

AI agents handle exploration, implementation, refactoring, research and repetitive engineering tasks.

They are not treated as authorities.

Generated code is reviewed, tested and validated before becoming production code.

Architecture still matters

AI makes implementation faster.

It does not remove the need for architecture, system design, debugging, performance analysis or engineering judgment.

WHAT I'M LOOKING FOR

I'm interested in remote teams building ambitious products where AI-native engineering is part of the actual development process.

Particularly interested in:

  • AI-native products
  • AI developer infrastructure
  • Agentic systems
  • MCP infrastructure
  • Developer tools
  • Product engineering
  • Web3 / Crypto
  • Mobile & cross-platform software
  • Networking & real-time systems
  • Experimental startups
  • Small autonomous engineering teams

TARGET ROLES

Senior Product Engineer

Staff Product Engineer

AI-Native Software Engineer

AI Product Engineer

Founding Engineer

Developer Tools Engineer

Senior Mobile Engineer

TECHNICAL PROFILE

Primary

Swift · iOS · macOS · React Native

Systems

Rust · Networking · Streaming · Real-time Communication · Media · Background Services · CLI

AI Engineering

MCP · AI Agents · Agentic Development · Kiro Spec-Driven Development · Oh My Pi · Semantic Search · LLM Applications

AI Infrastructure

PostgreSQL · pgvector · Embeddings · Vector Search · Custom MCP Servers · Persistent Knowledge Systems

Development

Zed · Git · GitHub · Git Worktrees · CI/CD · Bash · Automation

Engineering Style

Product-oriented · Autonomous · Systems Thinking · Specification-driven · AI-native