Code examples and quickstarts for the Zylora serverless GPU platform.
| Language | Package | Docs |
|---|---|---|
| Python | zylora |
Quickstart → |
| TypeScript / Node.js | @zylora/sdk |
Coming soon |
Write a normal Python function, add one decorator, deploy — done.
fn/
01_embed.py Sentence embeddings on a T4
02_image_caption.py Image → caption on an A100
03_llm_stream.py Token streaming from Mistral-7B on an H100
04_batch_map.py Parallel .map() over a list of inputs
05_background.py Fire-and-forget .remote_async() + async fan-out
06_always_warm.py Keep instances hot (min_instances > 0)
07_latency_routing.py cost / latency / reliability routing strategies
Jump to → fn/README.md
Lower-level Zylora() client — useful when calling functions you don't own
or when integrating Zylora into an existing codebase.
python/
01_hello_gpu.py Basic sync invocation
02_async_invoke.py Async invocation with asyncio
03_streaming.py Token-by-token SSE streaming
04_batch.py Parallel batch invocation
05_background_job.py Fire-and-forget job with polling
Jump to → python/README.md
All examples assume you have a Zylora account and an API key:
export ZYLORA_API_KEY=zyl_...The fn/ examples also require a deployed function — run zy deploy <file> first.
Sign up at zylora.dev.