AI - Arnold Biffna Portfolio

Practical AI Integration for Apps, Automation, and Developer Productivity

I build practical AI features into real applications and workflows — from on-device Apple Intelligence in educational apps to OpenAI-powered automation, local AI media processing, AI-assisted development, and internal tools that reduce repetitive work.

My focus is not AI research. My focus is applied AI: identifying useful problems, choosing the right tools, integrating them into working software, and using AI to improve real products, workflows, and developer productivity.

What I Do with AI

App Features

Apple Intelligence integration
OpenAI API integration
AI-generated educational content
Cover letter and document generation
AI-powered descriptions, summaries, and metadata

Developer Productivity

Claude Code, OpenAI Codex, ChatGPT, and Grok-assisted development
Architecture planning and refactoring
Debugging, testing, and code migration
MCP services and local workflow automation
Rapid prototyping across Swift, Kotlin, Python, C#, and .NET

Media and Automation

Local AI image and video descriptions
Keyword and tag generation
Subtitle generation
Batch media processing
AI-assisted video, script, and social content creation

Featured AI Work

TimesX — Apple Intelligence Word Problems

TimesX is an educational multiplication practice app available on the App Store and Google Play.

I rewrote TimesX in native Swift and integrated Apple Intelligence to generate multiplication-based word problems on device. This gives students fresh practice content while preserving privacy and allowing the app to work offline on supported devices.

AI used for: on-device educational word problem generation
Technologies: Swift, Apple Intelligence, Kotlin, Android, Chromebook support
Related article: TimesX 2026, now with AI

Local AI Media Processing Pipeline

I built a local AI workflow to generate searchable descriptions and tags for a large personal archive of roughly 30,000 photos and videos. The system uses a Python worker, MySQL, FFmpeg, ExifTool, and a local oMLX server running on Apple Silicon to process media without paying per cloud API request.

The workflow separates the job runner from the AI server: an Ubuntu machine manages the Python processing pipeline while a Mac runs the local AI model. Images are resized before analysis, videos are sampled with FFmpeg, and results are stored in MySQL so the process can safely resume if interrupted.

AI used for: image descriptions, video frame descriptions, searchable metadata, keyword generation
Technologies: Python, MySQL, FFmpeg, ExifTool, oMLX, Apple Silicon, Ubuntu
Related article: Local AI is Saving Me $

JobHunt — AI Cover Letter Generation

JobHunt is a personal macOS application designed to organize job opportunities and streamline the application process. It uses a résumé and job description as context to generate customized cover letters through a REST API connected to OpenAI.

This project combines desktop productivity software, AI content generation, and workflow automation around a real job-search use case.

AI used for: cover letter generation, résumé/job-description context, writing assistance
Technologies: macOS, OpenAI API, REST API, Cloudflare Workers

AI-Assisted Software Development

Alien Barrage — AI-Assisted iOS Game Development

Alien Barrage is a retro-inspired iOS shoot-’em-up game built with Swift and SpriteKit. It includes Game Center leaderboards, achievements, in-app purchases, localization, powerups, bosses, and multiple gameplay systems.

Claude and Codex helped accelerate development, planning, localization, documentation, asset workflows, debugging, and repetitive implementation tasks. The AI tools were valuable, but the project still required architecture decisions, testing, platform integration, gameplay tuning, and release management.

AI used for: code generation, architecture planning, debugging, localization, documentation, asset workflows
Technologies: Swift, SpriteKit, Game Center, StoreKit, Claude, Codex
App Store: Alien Barrage
Related article: Alien Barrage — Building an iOS Game with AI

TimesX Native iOS Migration

I migrated TimesX from Xamarin iOS in C# to native Swift. AI-assisted planning helped break the migration into phases, evaluate architecture decisions, and accelerate implementation while improving the app’s design and user experience.

AI used for: migration planning, Swift implementation help, refactoring, debugging
Technologies: Swift, Xamarin, C#, Apple Intelligence

App Store: TimesX Times Table Practice

TimesX Android and Chromebook

Using lessons from the iOS migration, I built a native Kotlin version for Android and Chromebook. AI-assisted development helped speed up implementation while I focused on platform-specific behavior, landscape layouts, and classroom-friendly usability.

AI used for: Kotlin implementation support, UI iteration, testing assistance
Technologies: Kotlin, Android, Chromebook, Google Play
Google Play: TimesX Times Table Practice

MCP Services and Workflow Automation

I have implemented Model Context Protocol services to extend AI coding assistants such as Claude and Codex. These services expose local command-line tools, project workflows, and automation tasks across my Mac and Ubuntu systems.

AI used for: local tool access, automation, development workflow extensions
Technologies: MCP, Python, command-line tools, macOS, Ubuntu

AI Media and Automation Experiments

Plain-English FFmpeg Automation with Claude

FFmpeg is powerful, but complex. I used Claude as a plain-English interface for FFmpeg, turning a video editing goal into a finished Instagram-ready MP4 with background video, centered app footage, speed changes, padding, and QR code overlay.

AI used for: command generation, media workflow planning, automation assistance
Technologies: Claude, FFmpeg, video processing
Related article: Advanced FFmpeg in plain English using Claude

AI Subtitle Generation

I explored using AI transcription tools to generate subtitles for video content, including multilingual media workflows. This connects to my broader interest in making video and audio archives easier to search, understand, and reuse.

AI used for: transcription, subtitle generation, multilingual media support
Technologies: AI transcription tools, video processing
Related article: Generate Subtitles for Your Videos Free with AI

Local AI Voice and Audio Experiments

I experimented with a local Python-based voice and audio workflow using FFmpeg, MLX Whisper, MLX Audio, and Hugging Face model snapshots. The project helped me explore how local AI can support transcription, reference-audio cleanup, and text-to-speech generation without relying on cloud services.

I treat this as an AI audio experiment rather than a production product. It is useful for understanding local AI capabilities, audio preprocessing, transcription, and responsible synthetic media workflows.

AI used for: transcription, reference audio cleanup, text-to-speech experimentation
Technologies: Python, FFmpeg, FFprobe, MLX Whisper, MLX Audio, Hugging Face
Related article: Voice Cloning

Internal Tools and Automation

Biff Download

Biff Download is a personal Apple TV and home-network media system using Python services for content discovery, downloading, subtitle management, file organization, and AI-powered subtitle generation.

AI used for: subtitle generation, media organization, automation
Technologies: Python, Apple TV, local media workflows

JobScrapers

JobScrapers is a Python-based job scraping platform that collects job listings from multiple sources using Playwright and Selenium. It stores results in MySQL and uses AI-assisted categorization and filtering to help identify opportunities that match specific interests and skills.

AI used for: categorization, filtering, workflow assistance
Technologies: Python, Playwright, Selenium, MySQL, AI-assisted classification

S3 Uploader

S3 Uploader is a desktop utility that simplifies uploading files to Amazon S3. It is an example of using AI-assisted coding workflows to quickly build focused internal productivity tools.

AI used for: rapid development, implementation support, debugging
Technologies: desktop app development, Amazon S3, AI-assisted coding

Technologies Used

AI Models and Services

Apple Intelligence
OpenAI GPT models
Claude
Grok
Qwen Coder
MLX Whisper
MLX Audio
Local oMLX models

AI Development Tools

Claude Code
OpenAI Codex
ChatGPT
MCP

Languages and Platforms

Swift
Kotlin
Python
C#
.NET MAUI
REST APIs
Cloudflare Workers
MySQL
macOS
Ubuntu
Apple TV

Use Cases

Content generation
Educational question generation
Cover letter generation
Media description and tagging
Subtitle generation
Local AI processing
Workflow automation
AI-assisted software development
AI-assisted media production