Olympus.
Platform to control AWS cloud resources (S3, EC2, Lambda) through natural language via NVIDIA's Nemotron LLM, MCP, and Terraform.
Olympus lets you manage cloud infrastructure by typing plain English commands. A React frontend talks to a Flask backend, which forwards requests to a Node.js MCP client, which executes Terraform inside a Docker container to provision or destroy AWS resources. NVIDIA's Nemotron Nano LLM routes natural language to the right Terraform tool (createS3Bucket, createEC2, createLambda, etc.). Per-resource mutex locks prevent race conditions on Terraform state. I also built an automated ticketing system that assigns incoming issues to users based on their skills.
Problem
Provisioning cloud resources still requires writing Terraform configs or clicking through the AWS console. There's no natural language layer that maps plain intent to the right infrastructure operation.
Solution
Route natural language through NVIDIA Nemotron → Node MCP client → Dockerized Terraform server. Each resource type (S3/EC2/Lambda) gets its own mutex so concurrent requests can't corrupt Terraform state.
- 01Natural language commands routed through NVIDIA Nemotron Nano to 7 Terraform MCP tools
- 02Provisions and destroys S3 buckets, EC2 instances, and Lambda functions via a Dockerized Terraform server
- 03Per-resource mutex locks (S3/EC2/Lambda) prevent concurrent Terraform state corruption
- 04Automated ticketing system assigns issues to team members by skill match
- 3AWS Services
- 7MCP Tools
frontend
- ReactJS
backend
- Flask
- Node.js
database
- DynamoDB
deployment
- Docker
- AWS (S3, EC2, Lambda)
other
- Terraform
- NVIDIA Nemotron
- MCP (Model Context Protocol)