04.11.2026 aktualisiert


verifiziert
Premiumkunde
100 % verfügbarAI Engineer | Senior Apple Platform Engineer | iOS, Swift, OpenAI Agents SDK, Codex
Berlin, Deutschland
Weltweit
General Computer Science, Charles University - Czechia (2007 - 2009)Über mich
AI Engineer and Senior Apple Platform Engineer with 15+ years shipping software. I build AI agents and developer tooling in Python and TypeScript with OpenAI, Codex and MCP, alongside iOS/macOS delivery in Swift. Banking experience, tested workflows and human-reviewed execution.
Skills
AI AgentsRAGLLMOpenAIChatGPTIpad AppArchitekturUnit TestingBankingMobile Application DevelopmentCode-ReviewConcurrent ComputingContinuous DeliveryContinuous IntegrationDebugging
AI Engineer and Senior Apple Platform Engineer with 15+ years shipping production iOS, iPadOS, macOS and fintech/mobile banking software. I combine Apple platform expertise with applied AI engineering: OpenAI, Codex, AI agents, agentic workflows, LLM applications and developer tooling in Python and TypeScript.
Apple Platform / iOS AI: Swift 6, strict concurrency, async/await, actors, Sendable, SwiftUI, UIKit, AppKit, iOS, iPadOS, macOS, tvOS, Observation, SwiftData, Swift Charts, Metal, Kotlin Multiplatform integration and DTO boundaries. Local AI exploration: Ollama, Core ML, ML Kit, on-device AI, local inference and privacy-first AI integration.
OpenAI / Agentic AI: OpenAI API, OpenAI Agents SDK, Codex SDK, MCP/connectors, tool calling, prompt engineering, instruction design, context engineering, workflow orchestration, handoffs, human-in-the-loop review, safe tool execution and verification loops. API/connector-based retrieval-augmented generation (RAG): source-grounded requirements synthesis from Figma, Jira and Slack.
Personal AI tooling: Agentic Review (TypeScript, XState, Codex SDK) combines GitHub/Jira context, CodeRabbit findings and Codex verification, with resumable state, schema validation, revision checks and human approval before GitHub writes. Relay (Python, Codex) coordinates long-running planner/executor/reviewer workflows with persistent handoffs, recovery and revision-bound review evidence. These are tools built for my own engineering work.
Codex / AI-assisted Delivery: repo-local workflows with Codex, Claude, CLI agents, skills, plugins, MCP tools, worktrees, Plan/Goal Mode, subagents, parallel tasks, code review, refactoring, migration planning, test generation, debugging, documentation and CI verification. Focused on auditable, reproducible changes in real codebases.
Architecture & Quality: Clean Architecture, MVVM+C, Coordinators, Clean Swift/VIP, VIPER, modular architecture, SwiftPM, XCTest, Swift Testing, pytest, Vitest, unit/UI/snapshot/integration tests, CI/CD, GitHub/GitLab, secure configuration, authentication/session lifecycle and Instruments. Languages: Python, TypeScript/JavaScript, Swift, Objective-C, C/C++, Bash; data work with NumPy, Polars and PyArrow.
Domain Context: regulated fintech and mobile banking, including Deutsche Bank, Postbank, FYRST and BestSign. Bringing production architecture, concurrency, security and testing experience to AI-native Apple apps, OpenAI/LLM integrations and reliable agentic workflows.
Erfahrung
Berufserfahrungmehr als 10 Jahre
Freelancer seit2009
Sprachen
DeutschgutEnglischverhandlungssicherTschechischMuttersprache
Projekthistorie
Apple-platform client delivery alongside applied AI engineering and developer tooling for my own work.
Client delivery: iOS, iPadOS, macOS and tvOS; Swift 6 migrations, strict concurrency, async/await, actors, Sendable and data-race fixes; SwiftUI/Observation, DTO boundaries, Kotlin Multiplatform integration, Swift Charts/Metal, tests and CI/CD. Integrated Codex and Claude into implementation, refactoring, documentation and code review with build/test verification.
Personal AI tooling:
- Agentic Review (TypeScript, XState, Codex SDK): PR review runner combining GitHub/Jira context, CodeRabbit findings and Codex verification; resumable state, schema validation, revision checks, human approval before GitHub writes and failure-path tests.
- Relay (Python, Codex): long-running planner/executor/reviewer workflows, persistent handoffs, recovery and validation of revision-bound review evidence.
- Retrieval-backed requirements synthesis: reusable skills combining Figma, Jira and Slack sources into technical specifications; API/connector-based RAG, prompt/context engineering and source grounding.
Prototyped OpenAI Agents SDK and MCP integrations, tool calling and local AI with Ollama/on-device model experiments. Focus: reliable agentic workflows, clear data/tool boundaries and auditable, human-reviewed delivery.
Led development of greenfield, white-label mobile banking and signing apps for Deutsche Bank group brands including Postbank, FYRST and BestSign. Senior Apple platform / iOS engineering in a regulated fintech environment with high security, release quality and maintainability requirements.
Focus areas: Swift, native iOS, modular Clean Architecture, Coordinators, RxSwift, Swift Package Manager, secure configuration, authentication/session lifecycle, transaction verification, CI/CD, environment configuration, code review, mentoring and production releases.
This work is directly relevant to AI-native Apple apps and OpenAI/LLM features that need secure architecture, privacy-aware data boundaries, tool-driven delivery, testing, review loops and maintainable integration into existing enterprise mobile codebases.
Developed and maintained enterprise-grade iPad apps for Deutsche Bank corporate banking and branch workflows. Apple platform engineering for security-sensitive internal iOS/iPadOS applications with a focus on reliability, usability, performance and production delivery.
Focus areas: Swift, UIKit, iPadOS, Clean Swift/VIP architecture, enterprise app workflows, Apple Enterprise Design Labs collaboration, concurrency and data-race debugging, dependency modernization, Workspaces, unit/UI testing, code review and developer experience improvements.
Relevant for AI-native work where new AI, OpenAI, Codex or LLM-assisted features must fit into secure business-critical Apple workflows without breaking reliability or compliance expectations.