← Back to portfolio
omerfarukaydn.com
Uses · Updated August 2026
What I use to ship.
A complete list of the hardware, software, services, and tools I use daily. Inspired by uses.tech . If something here isn't linked, it's because there's no canonical source.
💻 Hardware
MacBook Pro 16" M3 Max
Main work machine. 64GB RAM, 1TB SSD. Handles 70B local LLMs surprisingly well via llama.cpp + Metal.
Custom PC (Linux)
AMD Ryzen 9 + 64GB RAM + RTX 4090. Where the heavy ML training happens. Runs Ubuntu 24.04 LTS.
LG UltraFine 5K
Primary monitor. 27", 5120×2880. Color-accurate, USB-C single-cable dock.
Logitech MX Master 3S
Best mouse for productivity. The horizontal scroll wheel is a game changer for timeline-based work.
HHKB Professional Hybrid Type-S
Mechanical keyboard, Topre switches. Compact 60% layout. Quiet enough for shared spaces.
Sony WH-1000XM5
Noise-cancelling headphones. Essential for focus work and conference calls.
🖥️ Editor & Terminal
VS Code fork with native AI integration. Claude 3.5 Sonnet in the sidebar, Cmd-K to edit, agent mode for multi-file changes. The single biggest productivity boost of 2024.
Backup when Cursor's context is too narrow. Also useful for boilerplate and tests.
Modern terminal with AI command search, block-based output, and proper input editing. Replaces iTerm2 for me.
Design tool. I design my own portfolio + marketing pages in Figma, then code them.
🛠️ Development Stack
App Router, RSC, Server Actions. My default for any web product. The portfolio itself is plain HTML/CSS/JS, but anything with auth or data uses Next.
Postgres + Auth + Realtime + Storage. The fastest way to ship a multi-tenant SaaS backend.
Python backend. My default for AI/ML serving — the OpenAPI docs are auto-generated and Pydantic validation is bulletproof.
Desktop apps. Rust backend, web frontend. Used in Strata (local LLM manager). Replaces Electron for me — 10× smaller binaries, way faster startup.
Local LLM inference. The most important open-source AI project of the last 2 years. Run any GGUF model on consumer hardware.
Deployment for the portfolio and most web projects. Edge functions, automatic SSL, GitHub integration. Where this site lives.
🤖 AI Services
GPT-4o for production assistants, o1 for complex reasoning, text-embedding-3-small for retrieval.
Claude 3.5 Sonnet for code generation and structured extraction. Better than GPT-4 at following complex JSON schemas in my testing.
Multimodal work, and a fallback for cost-sensitive tasks. The 1.5 Flash model is great for high-volume, low-stakes pipelines.
Voice generation for ShortAI and other audio projects. 30+ voices, multi-language, emotion control.
📊 Data & Trading
Unified crypto exchange API. Binance, Coinbase, Kraken, Bybit — one interface. Used in the trading bot.
Time-series Postgres extension. Used for crypto price history and metrics.
Bayesian hyperparameter optimization. 18 minutes for XGBoost tuning vs 4 hours for grid search. Massive difference in iteration speed.
🌐 Hosting & Infrastructure
Portfolio, Next.js apps, serverless functions. Frankfurt edge. SSL auto-renewed.
Managed Postgres + Auth. Where Assistora AI lives.
Domain registrar. omerfarukaydn.com is registered here. Good support, Turkish billing.
DNS + DDoS protection for non-Vercel services.
📨 Communication & Productivity
Project management + personal wiki + specs + meeting notes. Everything lives here.
Issue tracking for client work. The keyboard shortcuts alone make it worth it.
Async communication with collaborators and clients.
Async video for client updates and code walkthroughs. Replaces most meetings.
Passwords, API keys, SSH keys, 2FA codes. Worth every penny.
The only calendar app that handles 3 timezones without making me want to throw my laptop.
🎵 Music for Focus
Generative soundscapes. The "Focus" preset is scientifically designed to keep you in flow state.
AI-composed music for deep work. The 40Hz gamma preset is the closest thing I've found to a productivity drug.
📚 Learning
Books, videos, live courses. Subscription pays for itself in 2 weeks.
Best Python architecture tutorials. Watched every video on his channel.
Deep-dive courses from real practitioners. The Next.js course by Scott Moss is gold.
Inspired by uses.tech . Last updated 20 August 2026.