At Golem, we're thrilled to introduce you to one of our exciting projects that has just reached a Pre-Alpha stage: Ray on Golem. On top of that, we are offering a unique collaboration opportunity for Python developers of all levels who want to try it out. Before we go into the specifics of this opportunity, let’s dive into the details of the project:
What is Ray on Golem?
Ray on Golem is an exciting integration with Ray, a distributed computing framework, to provide an easier way of accessing the Golem Network for Python developers. It can also be a powerful and cost-effective infrastructure option for Ray users. By leveraging the decentralized computing resources of Golem Network, Ray on Golem aims to enable the seamless execution of Ray applications and address performance limitations associated with traditional infrastructure providers. 🔁⚡️
Why are we doing Ray on Golem?
The primary motivation behind Ray on Golem is to make it easier to run Python computations on Golem. As a side effect, we want to enhance the perception and utility of Golem by showcasing its compatibility with Ray and its potential to address real-world use cases. Through collaboration between the Golem team, Ray contributors, and Golem community developers, we seek to validate hypotheses, gather feedback, and make informed decisions regarding the development and implementation of Ray on Golem. Additionally, we aim to bridge connections with MLOps teams to explore collaboration opportunities and provide infrastructure options for companies seeking alternative solutions. 🚀🤝🌍
How are we doing this project?
Ray on Golem follows a phased approach to achieve its goals:
In Phase 1, we conduct research to understand the challenges faced by Ray users on AWS and evaluate the viability of Ray on Golem from a user perspective. We establish contact with Ray users to evaluate the project idea and gather valuable insights. 🔬📝
In Phase 2, we focus on building the Minimum Viable Product (MVP) to offer a ready-to-use solution for executing Ray applications on the Golem Network. Throughout this phase, we actively seek feedback on the user experience (UX) to refine and improve the offering. 🛠️💡 Currently we are in this phase.
In Phase 3, we develop real-world use cases and actively engage with the community, including Ray contributors, data scientists, MLOps professionals, and Golem providers. This engagement allows us to gather feedback to fuel decision-making, assess the viability of Ray on Golem, and determine if further investment is justified. We also aim to obtain real-world benchmarks for Ray applications running on Golem to endorse our platform and enable Ray payloads on our mainnet. 🌍📊
Who is it for?
Ray on Golem targets several personas within the Ray and Golem communities. The personas include:
- Golem Community Devs: Developers within the Golem community will contribute with feedback, exploring CPU (and GPU later on, when Golem has this feature implemented & ready) Ray on Golem applications, and helping to improve the overall offering. They are the main target group - Ray on Golem is supposed to make it easier for them to run their payloads on the Golem Network and to decrease their dropout rate. 👩💻👨💻🌟
- Ray Contributors: These individuals will provide feedback on the UX, contribute real-world Ray on Golem use cases, and help integrate Golem into the Ray repository and documentation. They will also play a crucial role in overcoming performance issues. 👥📝🔧
- Ray App Devs - Data Scientists: Data scientists will be encouraged to try Ray on Golem (e.g. via Jupyter Notebooks) and contribute with real-world Ray on Golem applications. 🧪📊🔬
- MLOps: When we support GPU, MLOps professionals will be able to explore Ray on Golem as a cost-effective option and recommend Ray to data scientists within their organizations. 🤖💼💡
- Golem Providers: Experimental GPU providers will be invited to provide and utilize GPU resources, enabling GPU-specific use cases on the Golem network. 💻💡🔌
The integration of Ray on Golem offers a much better Golem developer UX while enhancing performance and scalability and enabling efficient execution of Ray applications. This collaboration expands the utility of Golem, making it a valuable infrastructure option for Ray users. Additionally, the project fosters community engagement, facilitates feedback-driven development, and validates the viability of Ray on Golem through real-world benchmarks and endorsements. 🔄🚀💪
You can explore the documentation of the Pre-Alpha version here: https://docs.golem.network/docs/creators/ray
Task Opportunity: Get Involved!
We are excited to offer a paid task opportunity to gather valuable feedback for the next steps of the project. We're on the hunt for 5-10 Python individuals to dive into Ray on Golem. All you need to do is follow our guide, test it out, and share your insights with us. Your feedback will be instrumental in shaping the project's future!
📋 What are the Requirements?
Only basic Python skills are needed, and previous experience with Ray is not required. 😃
💰 What's In It For You?
Spend just 1-2 hours of your time, and in return, we are offering a compensation of 600 GLMs.
📝 Ready to Jump In?
If you are interested in participating in this exciting opportunity, don't wait! Fill out the form right here: https://bit.ly/48j50ap by Wednesday, September 27. Once you have filled it, we will come back to you with information about the next steps.
We're excited about the potential of Ray on Golem and we invite you to join us on this incredible journey! 🌐👨🏻💻🌟
Stay tuned for more updates, and let's shape the future of Golem Network together! 🌍🔗✨