What I run

MacBook

Current daily driver

NVIDIA DGX Spark

GB10 — unified memory, runs Lumbridge Compute

GPU workstation

Local training & inference

Prime Intellect GPUs

On-demand SFT and evaluation runs

Cloud compute

Multi-region, always-on services

Next.js

App Router, React

React + Vite

SPAs, dashboards, internal tools

TypeScript

Strict mode, always

Tailwind CSS

Utility-first styling

shadcn/ui

Component primitives

WebGL

Custom shaders (Three.js raw)

Tauri

Rust-powered desktop apps

Electron

Cross-platform desktop

Swift

iOS / macOS native

Kotlin

Android / JVM

Node.js

Bun runtime where possible

Fastify

Servers behind the karti.ai network

Express

Service APIs

Go

Services, MCP servers, CLIs

Python

Trading engine, data pipelines

WebSocket

Real-time data feeds, multiplayer rooms

LiveKit

Self-hosted voice, video & screen share

Event-driven engine

Rust + Python core

Real-time market data

Multi-venue feeds

Backtesting & research

Strategy simulation

PostgreSQL

Primary data store

SQLite

Local-first app data (better-sqlite3)

Redis

Caching, sessions, queues

Supabase

Managed Postgres edge, auth, storage

Prisma / Drizzle

Type-safe ORMs

Docker

All services containerized

Caddy

Reverse proxy, auto HTTPS

Proxmox

Homelab virtualization

Coolify

Self-hosted PaaS

Dokploy

Self-hosted app deploys

GitHub Actions

CI/CD pipelines

Cloud

ARM Ampere A1 instances

Tailscale

Private mesh VPN

WireGuard

VPN tunnels

Buzz

Self-hosted team comms on a Nostr relay

Claude Code

Agentic coding

Codex

OpenAI coding agent

Devin

Autonomous SWE agent

Cursor

AI-native editor

Orca IDE

Parallel-agent IDE (worktrees)

Opencode

Custom AI workspace

cmux

Terminal for parallel agents

Oh My OpenAgent

Multi-agent orchestration

Archon

System architecture engine

Custom orchestration

Multi-agent workflows

Prime Agent

Tool-using agent harness

Background runners

Scheduled + autonomous tasks

Transformers + TRL

Model loading and supervised fine-tuning

vLLM

Local and hosted model serving

Prime Intellect Verifiers

Environment and held-out agent evaluation

LM Studio

Local LLM inference

MLX

Apple Silicon ML framework

Llama.cpp

Local LLM runtime

Fireworks AI

Serverless LLM inference

Together AI

Inference & fine-tuning

Deep Infra

Serverless model inference

Prime Intellect

On-demand GPU training and evaluation

TypeScript

Daily driver

Python

Data, ML, trading

Go

Services, CLI tools

Rust

Trading engine, systems programming

SQL

Query optimization

MCP servers

Built in Go, Python & TS

Infra & code

caddy, gitea, search

Comms

slack

Slack CLI

Workspace automation

gog CLI

Custom tooling

ripgrep

Code search

tmux

Terminal multiplexing