Spanish

🚀 Stream-Ready AWS Lambda with Pulumi & LangChain (TypeScript)

Deploy and test a streaming-capable Lambda function using modern IaC tools and AI workflows. Built with cutting-edge AWS features and LLM integration!

🌟 Features

• Lambda Response Streaming (AWS April 2023 feature) 💨
• Pulumi Infrastructure-as-Code (TypeScript flavor) 🏗️
• LangChain LCEL workflows integrated with AWS Bedrock (DeepSeek R1 model) ⚡
• Full-stack testing (Vitest + Pulumi Automation API) 🧪
• Zero-dependency deployment with pnpm 📦

🛠️ Prerequisites

• AWS Account (Free Tier OK)
• Node.js 18+
• pnpm
• AWS CLI configured with valid credentials

⚡ Quick Start

# Clone & setup
git clone https://github.com/renato1010/streaming-lambda-pulumi
cd streaming-lambda-pulumi
pnpm install

# Deploy infrastructure
pulumi up -y

# Run unit tests
pnpm vitest -t utils

# Execute integration tests
pnpm vitest -t streaming

🔍 Testing Architecture

1. Unit Tests (/tests/utils.test.ts)

// Example Vitest case
const base64String = 'eyJxdWVzdGlvbiI6IldoYXQncyBHZW5lcmF0aXZlIEFJLCBpbiAyMDAgY2hhcnMgb3IgbGVzcyJ9';
const expectedJsonString = `{"question":"What's Generative AI, in 200 chars or less"}`;

it('should decode base64', () => {
  const jsonString = decodeBase64String(base64String, true);
  expect(jsonString).toEqual(expectedJsonString);
});
it('should validate body payload against schema', () => {
  const bodyString = expectedJsonString;
  const validatedBody = queryBodySchemaValidator(bodyString, queryHandlerSchema);
  expect(validatedBody).toStrictEqual(JSON.parse(expectedJsonString));
});

2. Infrastructure Tests (/tests/streaming-function.test.ts)

⚠️: Here you are providing real infrstructure, so be generous with the timeouts

// Pulumi Automation API integration test
it('should have a streaming URL', async () => {
  const outputs = await getOutputs();
  const streamingUrl: string = outputs['streamingURL'].value;
  expect(streamingUrl).toBeDefined();
});
it('should return a response stream', { timeout: 60000 }, async () => {
  const outputs = await getOutputs();
  const streamingUrl: string = outputs['streamingURL'].value;
  const response = await fetch(streamingUrl, {
    method: 'POST',
    headers: {
      'Content-Type': 'application/json',
      Accept: 'text/event-stream'
    },
    body: JSON.stringify({
      question: "What's Generative AI, in 200 chars or less"
    })
  });
  expect(response.status).toBe(200);
  expect(response.headers.get('Content-Type')).toBe('text/event-stream');
});

⚠️ Warning: When running infrastructure tests, your stack is deployed and then destroyed. After running tests like those in tests/streaming-function.test.ts, you'll need to redeploy your infrastructure by running:

pulumi up

🎬 Live Demo: Stream Responses with HTTPie

http --stream POST $(pulumi stack output streamingURL )  \
  question="What's Generative AI, in 200 chars or less"

🧩 Tech Stack

Component	Role
🐑 Pulumi	IaC orchestration
⚡ AWS Lambda Streaming	Chunked response delivery
🔗 LangChain LCEL(TS)	Langchain Expression Language(Runnable)
🤖 AWS Bedrock	DeepSeek R1 model
🧪 Vitest	Unit/component testing
🤖 Pulumi Automation API	Infrastructure validation

🔗 Resources

• AWS Lambda Streaming Announcement
• Pulumi Streaming Lambda Guide
• LangChain JS Documentation

Arch Diagram

---

Found this useful? ⭐ Star the repo _License: MIT 📜