Skip to main content
WAVIS supports three installation paths. Pick the one that matches your stack.

​
Python SDK

The Python SDK ships as a pre-compiled wheel — no Rust toolchain needed for the common case. 
pip install wavis-fhe
Requirements:
  • Python 3.9+
  • Linux (x86_64), macOS (Apple Silicon + Intel), or Windows (x86_64)
  • ~50 MB disk space
Verify the install: 
import wavis_fhe as wv
keys = wv.keygen()
ct = keys.encrypt(True)
assert keys.decrypt(ct) is True
print("WAVIS works!")

​
Optional: GPU acceleration (BYO GPU)

For local GPU acceleration in your own environment: 
pip install wavis-fhe[cuda]
Requires: NVIDIA GPU with Compute Capability ≥ 7.5 (Turing/RTX 20 or later) and CUDA 11.8+. The CUDA kernel is compiled at runtime via NVRTC — no nvcc needed at install time.

​
Build from source

You only need this if you want to modify the Rust core. 
# Install Rust toolchain
curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh

# Install maturin
pip install maturin

# Build the wheel
git clone https://github.com/ruka0911/wavis-code.git
cd wavis-code/python
maturin develop --release          # CPU
maturin develop --release --features cuda  # GPU

​
TypeScript / Node.js SDK

npm install @wavis/fhe-client
# or
yarn add @wavis/fhe-client
# or
pnpm add @wavis/fhe-client
Requirements:
  • Node.js 18+ or Deno 1.30+ or Bun 1.0+
  • Works in browsers (when bundled) and Cloudflare Workers
Verify the install: 
import { WavisClient } from "@wavis/fhe-client";

const client = new WavisClient({ apiKey: "wvs_trial_..." });
const health = await client.health.check();
console.log(health.status); // "healthy"

​
REST API (no install)

If you don’t want any local SDK, you can call WAVIS directly over HTTP. 
# Get a 30-minute trial key — no signup required
curl -X POST https://api.wavis.xyz/api/v1/onboarding/temp-key

# Health check
curl https://api.wavis.xyz/api/v1/health
See the REST guide for the full request/response shape.

​
Choosing a path

Python SDK

Best for: ML pipelines, data science, anything where you want gates locally.

TypeScript SDK

Best for: web apps, Node backends, browser-side encryption.

REST API

Best for: any language not listed above, or zero-dependency deploys.

​
Troubleshooting

​
error: Microsoft Visual C++ 14.0 is required (Windows)

You’re trying to build from source. Either install the Visual Studio Build Tools or use pip install wavis-fhe (pre-compiled wheel, no compiler needed).

​
ImportError: libcuda.so.1: cannot open shared object file

You installed wavis-fhe[cuda] but no NVIDIA driver is present. Either install the driver or use the CPU build: 
pip uninstall wavis-fhe
pip install wavis-fhe  # CPU only

​
keygen() is slow on the first call

First call compiles the bootstrapping key (~5 s on CPU, ~1 s on GPU). Subsequent calls are instant — keys are cached in memory.

​
Next Steps

Quickstart

Five lines of code, your first FHE gate

Authentication

API keys, environments, rotation