Alemno Payments

An asynchronous transactional data cleaning, validation, and LLM powered categorization pipeline.

Features

• Reads multi format dates and handles missing transaction IDs by auto generating unique fallbacks, and parses messy amounts with currency signs and commas cleanly.
• Offloads validation, spend conversion, and analysis to a Celery background worker backed by Redis.
• Integrates with the Google GenAI SDK (gemini-2.5-flash-lite 75% cheaper than `gemini-3.5-flash-lite with no tangible decrease in performance) using structured output schemas to categorize transaction records into database- defined custom categories.
• Uses Redis caching with a 24-hour TTL to store live exchange rates, falling back to a hardcoded rate of 93.0 if offline.
• Statistical Anomaly Detection:
  • Flags transactions exceeding 3 times the account's median spend as statistical outliers.
  • Flags domestic transactions in USD currency.
  • Flags high value transactions and transactions with fraud indicative keywords.

Click on the image to view the Draw.io file

Self Hosting & Deployment

We publish production-ready images to GitHub Container Registry (GHCR). You can pull them directly or run them via Docker Compose.

Create a docker-compose.yml file to host the entire stack (or just clone the repo and directly run docker compose up -d :

version: "3.8"

services:
  db:
    image: postgres:16-alpine
    container_name: alemno_db
    ports:
      - "5432:5432"
    environment:
      POSTGRES_USER: postgres
      POSTGRES_PASSWORD: postgres
      POSTGRES_DB: alemno_payments
    volumes:
      - postgres_data:/var/lib/postgresql/data
    healthcheck:
      test: ["CMD-SHELL", "pg_isready -U postgres -d alemno_payments"]
      interval: 5s
      timeout: 5s
      retries: 5

  redis:
    image: redis:7-alpine
    container_name: alemno_redis
    ports:
      - "6379:6379"

  web:
    image: ghcr.io/sortedcord/alemno-payments/web:latest
    container_name: alemno_web
    ports:
      - "3000:3000"
    environment:
      - POSTGRES_HOST=db
      - POSTGRES_PORT=5432
      - POSTGRES_USER=postgres
      - POSTGRES_PASSWORD=postgres
      - POSTGRES_DB=alemno_payments
      - REDIS_HOST=redis
      - REDIS_PORT=6379
      - REDIS_DB=0
      - GEMINI_API_KEY=your-gemini-api-key
    depends_on:
      db:
        condition: service_healthy
      redis:
        condition: service_started

  worker:
    image: ghcr.io/sortedcord/alemno-payments/worker:latest
    container_name: alemno_worker
    environment:
      - POSTGRES_HOST=db
      - POSTGRES_PORT=5432
      - POSTGRES_USER=postgres
      - POSTGRES_PASSWORD=postgres
      - POSTGRES_DB=alemno_payments
      - REDIS_HOST=redis
      - REDIS_PORT=6379
      - REDIS_DB=0
      - GEMINI_API_KEY=your-gemini-api-key
    depends_on:
      db:
        condition: service_healthy
      redis:
        condition: service_started

volumes:
  postgres_data:

Launch the stack:

docker compose up -d

Testing the API with curl

All API endpoints are prefixed with /api/v1. Below are copy-pasteable curl commands using the sample transactions.csv dataset.

1. Upload CSV Transaction Data

Submit a CSV file for asynchronous processing.

curl -X POST -F "file=@transactions.csv" http://localhost:3000/api/v1/jobs/upload

Response Example:

{
  "filename": "transactions.csv",
  "id": "c0249af1-c52d-4d49-9305-3d1a10f29e03",
  "status": "pending",
  "row_count_raw": null,
  "row_count_clean": null,
  "created_at": "2026-07-03T06:29:13.123456Z",
  "completed_at": null,
  "error_message": null
}

2. Check Job Status

Check the status of a job using the job ID returned by the upload endpoint.

curl http://localhost:3000/api/v1/jobs/c0249af1-c52d-4d49-9305-3d1a10f29e03/status

Response Example (while processing):

{
  "id": "c0249af1-c52d-4d49-9305-3d1a10f29e03",
  "status": "processing",
  "summary": null
}

3. Get Full Job Results

Retrieve the high-level summary along with all categorized and parsed transactions once the job is completed.

curl http://localhost:3000/api/v1/jobs/c0249af1-c52d-4d49-9305-3d1a10f29e03/results

4. List All Jobs

List all historical processing jobs. You can optionally filter by status (e.g. pending, processing, completed, failed).

# List all jobs
curl http://localhost:3000/api/v1/jobs

# List completed jobs only
curl "http://localhost:3000/api/v1/jobs?status=completed"

5. List Categories

List all available transaction categories sorted alphabetically.

curl http://localhost:3000/api/v1/categories

6. Create Custom Category

Add a new custom category. Newly uploaded transactions will dynamically map to this category during classification if identified by the LLM.

curl -X POST \
  -H "Content-Type: application/json" \
  -d '{"name": "Subscription"}' \
  http://localhost:3000/api/v1/categories