Are you a Master's/PhD/Undergrad student seeking an exciting project to work on and get paid?
Thinking about GSoC? That’s great, but do you know about RLOS? Let me introduce you to this program hosted by Microsoft every year! 👉
🤔 What is RLOS?
RLOS stands for Reinforcement Learning Open Source fest.
It's a 4-month program offered by Microsoft.
Accepted students receive a stipend of $10,000.
The program aims to invite diverse students worldwide to solve open-source RL problems.
🤓 Who am I?
Hey! I’m Nishant. I was one of the 10 people selected globally for this program in 2021. You will find me here on the alumni page.
I have a video on youtube demostrating my work at the end of the program. Do check it out! I’ve also written about it here.
💼 How does it work?
The usual process is as follows:
Select a project from their list of potential ideas.
Prepare and submit a proposal before the deadline.
You will be assigned a mentor upon selection.
Get paid by Microsoft.
Present your work to the world at the end of the program.
🌍 Similar to GSoC:
We all have heard about GSOC, right?? (If you havent, check out the video below!)
GSoC stands for Google Summer of Code. It is an opportunity to work on cool open-source projects with organizations, with payments ranging from $3000 to $6000 upon completion.
RLOS is very similar with some minor differences:
Direct collaboration with the Microsoft Research team in NYC.
Limited project choices, mostly centered around the machine learning library “Vowpal Wabbit”.
📢 By the end of the blog, you will know everything you need to, about RLOS.
🎓 Eligibility:
You need to be a student.
You need to be enrolled in a university or institute.
Masters, PhD, and Undergraduates are all eligible. For instance, I was a junior undergraduate when I got accepted. 📚
⏰ Timing:
Projects typically get announced around the last week of February. However, they may occasionally be delayed.
If you have inquiries, you can reach out to them at RLOSFEST@microsoft.com. They do respond! 📧
🔑 Prerequisites:
Understanding of basic machine learning concepts is essential.
Proficiency in basic Python or C++ is required.
Since you'll be collaborating on a codebase, it's important to be familiar with Git.
Familiarize yourself with the Vowpal Wabbit (VW) repository by brushing through their wiki.
Ability to self-learn any topic is crucial for success in the program. 📚
📝 Personal Tip: My project was based on multithreading. Before participating, I didnt know anything about multithreading, especially in c++. But while preparing the proposal, I taught myself whatever was essential. Google search and Youtube tutorials helped a lot! 🌟
Now with ChatGPT, learning a topic is even more easy and fast! 🚀
🔍 Types of Projects You Can Expect:
I encourage you to read through the listed projects for the last 4 years: 2023, 2022, 2021 and 2020.
There are mainly two types of projects that you may expect:
Improving the codebase: Most projects will focus on making VW better. By making it faster, adding more features, format conversions etc.
Research: Some projects are more focused on academia and try to test the limits of VW, create benchmarks or maybe work on a new algorithm.
📚 Past year demos:
I got to know about RLOS back in December 2020, when I got a suggestion on my youtube feed. Here’s the video:
At the end of the program each year, the participants come on a call and demostrate their summer’s work to the entire team. The recorded video is later upload on youtube.
Here’s the links for the 2023, 2022, 2021, 2020 part 1 and part 2.
🧠 Do you need to know RL?
Not necessarily. For example, the project that I worked on, had nothing to do with reinforcement learning. It was more on parallelizing a data ingestion pipeline used by VW and required the knowledge of multithreading.
However, there are some projects that are more research oriented, and require the knowledge of RL concepts. But you may skip those if you are not into RL.🚫
🎯Is there any preparation strategy?
Honestly, this is one of most requested questions I receive related to the program all over linkedin and whatsapp.
There are no specific preparation steps. Remember, this is not an internship or competition. Rather, it’s a way to spend your summers learning something interesting. So, just read through the listed projects and find what interests you.
That been said, I have tried some approaches that have helped me get selected in RLOS and GSoC.
For example, there are some steps you may follow while preparing your proposal to maximize your chance of getting selected. I’ll be sharing them along with some tips and tricks very soon here. Do subscribe to get notified.
🧐 More questions?
Go through their FAQ.
Still have questions? Paste them in the comments. I’ll try to answer as much as I can.
Cheers!🥂