Hardware-isolate,

New VS Code Extension

Your Local IDE.
Your Remote GPU.

Write ML code locally, sync in seconds, and execute on remote headless GPUs natively. No more SSH config headaches.

Or install via CLI
code --install-extension istidlal
gpu-extension — VS Code
Excellent Connection
< 50ms Latency
End-to-End Encrypted
Cloudflare Tunnels
Secure by Default
Zero Open Ports

Built for ML Engineers

Everything you need to develop ML models on remote GPUs as if they were local.

Real-Time Sync

Powered by Mutagen v0.18.1Bidirectional, low-latency syncing. Work locally in your preferred IDE while changes reflect on the remote GPU in milliseconds. One-way-safe mode prevents remote overwrites, keeping your local source sacred.
mutagen-sync-daemon
❯[Watch] Waiting for file changes...

Native Jupyter

Remote GPUs appear directly in your local VS Code notebook kernel picker. Run experiments locally and execute on high-performance remote clusters—zero SSH configuration required.

Live GPU Dashboard

Monitor VRAM utilization, temperature, and power draw in real-time. Track training jobs and catch Out-of-Memory (OOM) errors without ever leaving your editor.

Smart Auto-Recovery

Health monitors auto-pause file sync on connection failures to protect data integrity. The system automatically attempts reconnection and state restoration on the second heartbeat failure.

How It Works

Your local editor connects through a secure Cloudflare tunnel to Istidlal's GPU cloud.

Your Machine

VS Code

1. Local Development

Write code locally with your favorite editor and AI copilot

TUNNEL

2. Secure Tunnel

cloudflared creates encrypted connection to GPU server

Istidlal GPU Cloud

NVIDIA H100Slice 1/4
20 GB (MIG)
33%

3. Remote Execution

Code runs on GPU server with full local experience

Native VS Code Experience

Experience the power of remote GPU clusters with the simplicity of local development.

VS Code Extension Workflow Demonstration

vscodejupyterlab-hvddkx.istidlal.ai
Istidlal Extension
ISTIDLAL GPU DASHBOARD
🚀 Get Started
1Set up your GPU server (host, SSH key, etc.)
2Connect and start working
📙test.ipynb
# Save this as benchmark.py and run it: /opt/conda/bin/python benchmark.py
import socket
import sys
import torch
import time
device = torch.device('cuda' if torch.cuda.is_available() else 'cpu')
print(f"Using device: {device}")
IstidlalConnected: Tesla V100 GPU
Spaces: 4UTF-8Cell 3 of 3Prettier ✓

Pro Tip: Use the Cmd+Shift+I shortcut to toggle the GPU dashboard instantly.