🍳 RecipeGraph

RecipeGraph is a full-stack, cloud-native personal project built to learn and practice modern distributed software concepts.
It’s a small but ambitious platform that lets users create, store, and query recipes — powered by a graph-based backend and containerized infrastructure.

This project is designed as both a portfolio piece and a long-term learning exercise to explore real-world DevOps, backend, and AI-assisted data engineering patterns.

🧠 Concept

The system exposes two main capabilities:

1. Create and store recipes in a versioned database.
2. Query and explore recipes through a connected knowledge graph.

The backend will be structured as an MCP server with a FastAPI application.
Recipes will be persisted in PostgreSQL and represented in Neo4j as a semantic graph of ingredients, cuisines, and tags.

🧰 Tech Stack & Tools

🎯 Goals

• Learn to build, deploy, and manage a small distributed system from scratch.
• Gain hands-on experience with Kubernetes, Helm, and GitOps principles.
• Experiment with observability, metrics, and CI/CD pipelines.
• Explore AI-adjacent tools (LangGraph, knowledge graphs) in a practical context.

🗓️ Roadmap & Progress

Use this section to track progress — check off items as you complete them.
Each phase focuses on one concept area.

• Initialize repository and README
• Configure linting, formatting, typing, and testing tools
• Set up virtual environment and dependency manager

• Design relational schema for recipes and versions
• Implement FastAPI CRUD endpoints
• Add Pydantic models for validation
• Include normalization and tagging logic
• Write unit and integration tests

• Create MCP server exposing recipe tools (create_recipe, get_recipes, search_recipes)
• Share schemas between MCP and API layers
• Verify persistence and retrieval via MCP tools

• Introduce Neo4j as a graph backend
• Build service layer for graph upserts and lookups
• Create LangGraph pipeline for normalization → tagging → graph updates
• Add endpoint for graph exploration

• Dockerize API, MCP, and database services
• Add Docker Compose for local orchestration
• Verify consistent builds across systems

• Package services with Helm
• Deploy locally via MicroK8s or kind
• Validate inter-service connectivity and health

• Add Prometheus instrumentation to API
• Deploy Grafana and visualize key metrics
• Include Postgres and Neo4j exporters

• Set up GitHub Actions for linting, testing, and image builds
• Push images to GitHub Container Registry (GHCR)
• Enforce CI on pull requests

• Bootstrap Flux in local cluster
• Organize manifests by environment (dev, prod)
• Confirm automatic sync from GitHub to cluster

• Deploy project to a small remote VPS running k3s
• Connect Flux for remote GitOps
• Validate public ingress access

• Install minimal Istio profile
• Configure ingress gateway and mTLS
• Visualize traffic routing in Kiali or Grafana

• Create a minimal React + Vite UI
• Add recipe creation and browsing views
• Connect to FastAPI backend

• Define custom Prometheus metrics for recipes and latency
• Build Grafana dashboards for application insights
• Configure alerts for critical thresholds

• Add local Kafka or Redpanda
• Create a Flink job for trending ingredient aggregation
• Display trends in Grafana or the UI

• Implement health probes and readiness checks
• Add backup routines for Postgres and Neo4j
• Conduct lightweight load testing

• Implement synonym mapping for ingredient search
• Add “similar recipe” endpoint
• Integrate LangChain or LangGraph enrichment

• Create DEMO.md with usage guide and setup steps
• Add architecture diagram and screenshots
• Tag final stable release