ai-prop

ai-prop is a Rust-based statistical microservice that exposes statistical calculation logic via both REST (axum) and gRPC (tonic) APIs. It interacts with a PostgreSQL database via the ai-infra dependency for certain operations.

APIs

The service exposes the following endpoints (available on both REST and gRPC):

• /calculate_proportions
• /calculate_pooled_estimate
• /calculate_z_statistics

Ports

• REST (Axum): Port 3000 (by default) or configured via the PORT environment variable.
• gRPC (Tonic): Port 50051 (by default)

Live Deployment

The REST API is currently deployed on Render and is accessible at: https://ai-prop-service.onrender.com

Example Usage (REST)

curl -s -X POST https://ai-prop-service.onrender.com/calculate_proportions \
-H "Content-Type: application/json" \
-d '{"overall": {"m": 10.0, "n": 20.0}, "group1": {"m": 5.0, "n": 10.0}, "group2": {"m": 5.0, "n": 10.0}}'

Response:

{"p_population":0.5,"p1":0.5,"p2":0.5}

Binaries

This project contains two primary binaries:

• server: The main microservice providing the REST and gRPC endpoints (run by default with cargo run).
• divide_population_by_s: A utility binary for dividing population data into database partitions. (run with cargo run --bin divide_population_by_s).

Local Development Prerequisites

• Docker container running with the postgres image
• To connect to the database and check tables content: docker exec -it <container id> psql -U postgres

Infra Setup (via the ai-infra repo)

Run the following commands in the ai-infra repository:

• diesel migration run (to create the partitioned table and partitions)
• diesel migration revert (to use the runtime partitions from step 3 below)

Getting Started

1. Add a .env at the root of the infra project with: DATABASE_URL=postgres://usr:pwd@localhost:5432/name-postgres (change for the name of your PostgreSQL database and usr/pwd)
2. cargo build
3. cargo run --bin divide_population_by_s