Rakrup Orchestrator

💡 Inspiration

Modern software systems require orchestrating countless microservices, databases, scripts, and APIs. Too often, this demands writing complex, rigid boilerplate code or deploying heavy, high-maintenance infrastructure. We wanted to build a platform that combines the power of Infrastructure as Code (IaC) with the simplicity of a declarative configuration. Inspired by this vision, we created Rakrup—a sleek, event-driven orchestration platform that lets anyone build, schedule, and scale complex pipelines in seconds using simple YAML and visual building blocks.

⚙️ What it does

Rakrup is an all-in-one scheduled and event-driven orchestration engine. It allows developers to:

• Write Workflows in Simple YAML: Define task structures, variables, retries, and triggers using clean, declarative configs.
• Build Visually: Interactive, real-time DAG (Directed Acyclic Graph) visualization allows drag-and-drop workflow editing.
• Run Any Code: Built-in runner support to execute scripts in Python, Node.js, Shell, R, Go, and more.
• Leverage Generative AI: Integrated Gemini AI Assistant guides developers, suggests syntax corrections, and auto-completes YAML workflows.
• Enterprise Security: Configured with a robust Basic Authentication layer to safeguard administrative dashboards.

🛠️ How we built it

Rakrup's full-stack architecture is designed for maximum speed and minimal footprint:

• Backend: Engineered with a highly scalable Java & Micronaut microservice framework.
• Database & Queue: Powered by a lightweight H2 database relational engine to manage state, logs, and queue execution reliably.
• Frontend: A snappy, responsive dashboard crafted using Vue.js and Vite.
• AI Integration: Seamlessly integrated the Google Gemini API (gemini-2.5-flash) to offer intelligent, contextual pipeline building assistance.
• Deployment: Containerized via Docker and successfully hosted on Hugging Face Spaces for zero-friction online demo access.

🚧 Challenges we faced

Decoupling and securing a full-featured workflow engine to run seamlessly in serverless and containerized environments like Hugging Face Spaces presented significant hurdles. We faced storage persistence constraints and had to carefully configure lightweight SQLite/H2 strategies to avoid file locks. Furthermore, integrating a secure, non-intrusive Basic Authentication protocol without breaking real-time websocket API communication was a major breakthrough.

🎓 What we learned

We gained profound insights into building reactive microservices, managing localized databases in serverless architectures, and using declarative state machines. We also mastered fine-tuning and prompting Large Language Models to successfully output structured, syntax-valid YAML definitions for orchestration engines.

🔮 What's next for Rakrup

We plan to expand our serverless Task Runner ecosystem to execute heavy data workloads on public clouds, introduce granular Role-Based Access Control (RBAC), and expand the AI assistant's capacity to auto-heal failing workflow runs in real-time.

Built With

• docker
• git
• google-gemini-api
• gradle
• h2-database
• html5
• hugging-face-spaces
• java
• micronaut
• vite
• vue.js
• yaml