Fast, scalable NeoCloud for AI.

from medjed_sandbox import Sandbox

sandbox = Sandbox.create(
    image="python:3.12",
    internet=True,
    timeout="24h",
)

result = sandbox.run("""
    import pandas as pd
    df = pd.read_csv('/data/dataset.csv')
    df.describe()
""")

print(result.output)
# rows          48213.0
# mean_score       87.4

files = sandbox.fs.list("/data")
print(files)

sandbox.destroy()

import { Inference } from "@medjed/sdk";

const inference = new Inference({ apiKey: process.env.MEDJED_API_KEY });

const endpoint = await inference.deploy({
  model: "llama-3-70b",
  gpu: "H200",
  replicas: { min: 1, max: 8, scaleOnQueueDepth: 4 },
});

console.log(endpoint.url);
// https://llama-3-70b.endpoints.medjed.ai

const completion = await inference.complete({
  endpoint: endpoint.url,
  prompt: "Summarize the latest GPU market report.",
  maxTokens: 512,
});

console.log(completion.text);

from medjed_sdk import Metrics

metrics = Metrics(api_key="...")

fleet = metrics.fleet()

for gpu in fleet.instances:
    print(gpu.id, gpu.utilization, gpu.memory_used, gpu.temp)
# gpu-7f2a  98.4%  77.1GiB  71C
# gpu-9c01  92.7%  74.8GiB  69C

metrics.alert(
    metric="utilization",
    threshold=95,
    duration="5m",
    webhook="https://ops.example.com/hooks/medjed",
)