AI research workflow, shipped as production infrastructure

Agentic search with a real cloud deployment.

This demo runs on AWS ECS Fargate behind an Application Load Balancer, built from Docker, provisioned with Terraform, deployed via GitHub Actions, and wired to Groq for cloud LLM inference.

Runtime AWS ECS Fargate
Inference Groq API
Infra as Code Terraform
Deployment GitHub Actions
01

Search Interface

User enters a standard or deep research query from the live web app.

02

Agent Workflow

FastAPI orchestrates planning, tool use, memory, and background job execution.

03

Cloud Runtime

Requests flow through ALB into ECS Fargate tasks with observability and health checks.

04

LLM Provider Split

Local development can use Ollama, while cloud deployment uses Groq-backed inference.

Interactive Demo

Run the agent live

Use standard mode for interactive streaming, or switch to Deep mode for a background job.

Search stack: DuckDuckGo Wikipedia SearXNG ArXiv YouTube PubMed Gutenberg OpenStreetMap

System Metrics

Background Jobs