Installation Guide

This guide covers a development / staging installation of the full platform. For production hosting, see Deployment.

Prerequisites

The AI recommendation engine requires:

  • Python 3.10+

  • PostgreSQL 13+

  • Qdrant server 1.7+

  • Ollama with the bge-m3:latest embedding model

  • An OpenRouter API key (for Gemini access)

  • 8 GB RAM minimum (16 GB recommended)

  • 50 GB free disk for Qdrant storage

The web and mobile clients additionally require:

  • Node.js 20+ and a package manager (npm / pnpm)

  • For mobile builds: the Expo CLI and an Android / iOS toolchain or Expo Go

PostgreSQL

sudo apt update
sudo apt install postgresql postgresql-contrib
sudo -u postgres psql

CREATE DATABASE agent_memory;
CREATE USER oim_user WITH PASSWORD 'password';
GRANT ALL PRIVILEGES ON DATABASE agent_memory TO oim_user;

The full platform schema (users, beneficiaries, opportunities, recommendations, applications, audit logs, and more) can be loaded from OIM Postgres Data Model.sql. See Data Model.

Qdrant

docker pull qdrant/qdrant
docker run -p 6333:6333 -v ./qdrant_storage:/qdrant/storage qdrant/qdrant

Ollama

curl -fsSL https://ollama.com/install.sh | sh
ollama pull bge-m3:latest

AI Recommendation Engine

cd oim-ai-recommandation-engine-staging
python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
cp .env.example .env      # then edit values (see Configuration)
uvicorn main:app --host 0.0.0.0 --port 8000 --reload

Once running, the interactive Swagger UI is available at http://localhost:8000/docs.

Web Frontend and Partners Portal

Both web applications are Next.js projects with the same workflow:

cd oim-frontend-staging          # or oim-partners-front-staging
npm install
cp .env.example .env.local       # set NEXT_PUBLIC_API_URL etc.
npm run dev                      # development server on http://localhost:3000

Mobile Application

cd oim-mobile-main
npm install
# configure the API base URL in .env
npx expo start                   # opens the Expo dev tools

Use the Expo Go app or a configured emulator to run the application. Production builds are produced with EAS (see eas.json and Deployment).

Docker

The AI engine, frontend, and partners portal each ship a Dockerfile for containerised deployment. Build each image from its repository root, for example:

docker build -t oim/ai-engine ./oim-ai-recommandation-engine-staging
docker build -t oim/frontend ./oim-frontend-staging
docker build -t oim/partners ./oim-partners-front-staging