 Hi, this is Yoosupan Bhartiya and we are here at Open Source Summit in Vancouver. And today, once again, we have with us Ibrahim Adat, Vice President of the Strategic Programs, AI at the Linux Foundation. Ibrahim is great to have you back on the show in person. Thank you very much. My pleasure. Yeah, and you do so much there that we'll try to check all of what you do there. You've released a lot of reports and of course, you'll lead a lot of first there. But before we talk about everything else, I want to get an update on the whole initiative efforts that are going update on what is happening with LF, AI and data. LF, AI and data was established in the Linux Foundation about five years ago in May of 2018. And we started with the vision that there's actually a lot of interest and a lot of activities happening in open source relative to AI and data. And the goal was to bring in a lot of these efforts and provide them with a neutral environment that enables and fosters collaboration between the projects, between the projects and the companies and the communities. So we launched LF AI and data in 2018 with nine member companies. And we had a single project that we used to anchor and launch the initiative. And throughout the years, five years later, almost a month from today, it would be five years exactly, we have 55 member companies that are active members of the foundation. And we are up to 49 hosted projects, which is really incredible because for the past, I would say two and a half to three years, we have been adding one new project to the foundation every single month. And in fact, we can add more. However, of course, we have the limitation of resources. Otherwise, we can add many more projects on a monthly basis. And our community of developers grew from a handful of developers to almost 60,000 developers active across these different projects coming from over 500 organizations. So you can see the impact of hosting in a foundation onto a new environment under an open governance. That has really massive impact on the growth of the project. So it's been really phenomenal. And we expect to continue with this growth, especially today with all the emphasis in AI on open source and collaboration and accelerating innovation. And actually one interesting data point is when we look at the landscape that we manage that covers the top 340, 350 projects in the ecosystem. There's a little over one million line of code being created every single week. So there is absolutely no organization out there that can beat that pace of innovation and development. So my call of action for individuals and organizations that are listening to this video is, you know, if you are involved in open source AI and data projects, please come and talk to us. There's really massive potentials for collaboration for increased innovation by working together on all these different enabling blocks versus then you sitting and developing it yourself. So really very exciting times in AI and data. These are because we talk about JetGPT all the time these days. And there's so much discussion about bad AI, good AI, and everything else. Talk a bit about when we look at the interesting thing in the software industry is that the hype, you know, generally everything gets hyped. And after that, press finds the next shiny object and we move on. So when we are talking about this whole, you know, all these new, you know, buzzwords in the AI space, how much you're seeing is that it's being an evolution that has been happening for a while because there are two kinds of AI or technologies, one which is being actually used realistically in industry which is solving a lot of problems and one is just a buzzword. So realistically, where are we heading? So actually a very good question. So certainly there is a lot of hype, but also to counter that there are some very significant advancement in particular areas. So it really depends on the use case. So, you know, what kind of AI technologies you want to use to solve a given problem in your supply chain, for example, or in managing, you know, your customer support. For example, I work with an organization that have over a billion subscribers. And a lot of their concerns is how can we adopt different AI technologies that will help us minimize our efforts when it comes to managing customer inquiries? Or how can we manage the supply chain of our mobile phones being sent to like hundreds of millions of customers and so on? And I think at the end of the day, it's not creating technology for the sake of technology. It's more we have a problem and we need to solve it. So what kind of technology can we collaborate on creating together to minimize our costs and accelerate our innovation and address the given problem at hand, which will solve not just my problem, but it also solve your problem and many other problems that are very similar or within the domain. So despite the large hype since the beginning of the year of this year 2023, there are still very significant technological milestone being achieved. And a lot of it is not necessarily what users see when they interact with the product or service. A lot of it is kind of embedded in different layers that are invisible. So I would take, for example, the Linux kernel as an amazing example where, you know, the tablet you're using, my phone, you know, your infotainment system in a car, you know, there's a Linux kernel in all of these. It's an amazing piece of software, but it's not visible to anybody. So a lot of the innovation in AI is actually happening at these different layers that may not necessarily be visible to you, but it's there and it's being helpful. Thanks for talking about AI. Now I want to switch gears to Italy and talk about in last year at, I think, Dublin, you know, PyTorch Foundation was also announced. Give us an update on what's going on with the Foundation. So PyTorch and the PyTorch Foundation is like the most exciting thing you can be working on today. So we launched the Foundation in Dublin in the fall of last year and this is basically a massive commitment from MEDA, the founding company behind the PyTorch project and its core supporters in terms of the project that came together and decided to go even further in opening up the project and its governance, hosting it in the Foundation and providing you to hold for it to enable more companies, not just to use the project and rely on it, but also to contribute to it and feel that they have a path to become leaders in the project as they invest resources and they contribute to it. So we started and we launched the initiative or the PyTorch Foundation in the fall, I think, September of 2022. And since then, we've been doing a lot of work in the background. So for instance, we have established the operation of the Foundation within the LF we have hired two dedicated resources in addition to myself to be supportive staff for the Foundation. So now we have a dedicated head of marketing for the PyTorch project and we have a dedicated technical PM in support of the project and the Foundation. And we have a lot of efforts that are being in-flight in terms of transitioning the project assets from MEDA, the founder of the project, to the LF Foundation. And with that, we're almost done. I mean, there's like a few couple items that are pending, but pretty much we're there. And we kind of put in an agenda for what our focus would be in 2023. And of course beyond, but I think we're starting with 2023 where we have a lot of efforts being either announced or in the process of being announced in terms of supporting the developers in the project, increasing the number of developers contributing to the project, bringing in new developers, providing tools for better workflows. We have programs in relation to marketing, PR and communication. We started a training program. So we have in progress developer training that would be available via Linux platform and edX, similar, for example, to CNCF, which is really a great example to follow. We're also working on a certification program for PyTorch developers, enabling PyTorch on additional hardware and platforms. And one very key areas for PyTorch is its relationship with academia. So there's really a lot of emphasis from the project founders and the members of the PyTorch Foundation on deep involvement with academia and research. And we're working on programs that will be announcing very soon with a suspect. And I think the cherry on the ice cream, if you wish, is we are very close to announcing that PyTorch Foundation is open to accept new members. As you know, we started with the initial six founding members to give us kind of a chance to establish the foundation and its parameters and its operation. And we are finalizing the structure to welcome new members. So I think in about a month, a month and a half, we will see that announcement. And at that point, we will be able to welcome and onboard the new members of the foundation. And from there on, it's just basically more members, more technical projects and initiatives within the PyTorch project, more initiatives from marketing perspective, from training, from certification. And all of it is in support of growing the project and its community. And as you're talking about, your drive for more members, talk a bit about who are the exiting members and how they are involved with the project. When the PyTorch Foundation was established, we had six founding members that included Meta, AMD, Microsoft, Google, AWS, and NVIDIA. And the premise was instead of focusing on building a portfolio of 20, 30 companies, joining the foundation and announcing a foundation with so many members, we realized that it takes actually a lot of efforts and a lot of time to rally the companies to provide the quotes and agree on the press release and all the different logistics and signing the agreements and so on. So it takes a significant amount of time. So we opted for announcing the foundation with the initial six founding members and opted and left all the kind of organizational activities in relation to the foundation structure to a later point. So that we can be very fast to market in a sense we can announce the foundation right away without having to wait for all the logistical items to be executed. So now we're doing this interview here in Vancouver in May 2023 and I believe within a month, a month and a half or so, we will be announcing that we are ready to welcome new members. And of course, when we welcome new members, there's a specific onboarding process we have to go to basically in terms of online agreements, documents and bringing in the individuals from these different organizations, plugging them on the mailing list, on Zoom. So there's like a whole automation that needs to happen and we are actually in the process of implementing this. So today, the six founding members are involved in two ways. The first one is on the board of the foundation and the second is via the marketing committee. So about two months after starting the foundation, we established the marketing committee whose mission is really to promote the project via different marketing program that the foundation will establish. And each of these founding members designate a representative to this committee and we're seeing actually a lot of documentation being created, we're seeing, for example, yesterday on May 9, we had a Pytorch mini summit. We're announcing, actually we already announced the Pytorch conference that will take place in October in San Francisco. We have multiple Pytorch mini summits happening throughout the year. So there's a lot of activities and all of that really is driven by the Pytorch foundation as in support of the project itself. Of course, we are at an event so I do want to ask about, if you have any plans to have any events also for the foundation or the project? Yes, so the anchor project or the main flagship project event is the Pytorch conference. So last year we had it in New Orleans in December, actually December 2nd. And this year we are holding the event in San Francisco in October, I think October 16, 17. And that is really the event that brings in together data scientists using Pytorch, developers contributing to the project, organizations using the project, bringing them together under the Sambrola exchange technical advancement, exchange best practices and exchange ideas on how to continue and flourish the technical project itself. So if you are using, contributing, deploying Pytorch, this is really a must attend event in this respect. You also work on a lot of reports that keep coming out. So talk a bit about any major reports or research work that you did in 2023, some key findings there. I like to write and actually this is a part of, you know, I completed a PhD in computer science and part of that training is learning how to write. I mean, you know, you need to write a thesis that's like maybe 300 pages and before that your master thesis, that's another, you know, so basically over time, you start using writing as a way to, you know, to exchange information and to pass knowledge. And at the next foundation, I'm very involved with the Linux foundation research team. So on a yearly basis, I publish maybe anywhere between three to five different reports. So the last couple of reports I wrote this year, one of them was focused on OSPOS, open source program offices and the other one was focused on GitHub best practices. And I will start with the second one. So in LFA and data, we host today 49 projects and all of these projects come to us and part of the onboarding into the foundation, we recommend a number of activities that the project has to implement or execute so that they become eligible for hosting with us. And some of these activities or tasks are related to GitHub. So a lot of projects come to us, they have, for example, poor documentation on GitHub. They don't have security measures put in place. They have some of their settings are not fine-tuned. They don't have a DCO running on the code. So there's really a lot of improvements that can be implemented. And the premise first was, you know, I'm gonna write a two-pager that every time we have a project coming into the foundation, I give them this page and I tell them, hey, you know, I'd like you to go and implement these different measures to improve your presence on GitHub. And over time, you know, for the past two years, I've been adding to it more tasks and more tasks. And eventually it came to be like four or five pages and I'm thinking, hey, you know, this might be a great idea for a report. So I took that and I expanded on it. And now we have like a 30-page report that's really anyone, you don't have to be hosting your project for the foundation. If you have a project on GitHub and you wanna make sure that you improve the security of the project, you improve the documentation, you improve the visibility of the project, you have better settings, you know, et cetera, I would encourage you to download that report and view it and really actually go and implement the recommendations. It will just make your life in managing the project a lot better and easier. So this is kind of one of the examples. And the other aspect or the other report is the open source program office. And again, you know, all of my publications are driven from practical experience, right? From being a practitioner. So I've created the Ospo at Samsung Electronics, I participated in the Motorola Ospo, I created the RXEN Ospo back in the early 2000, participated in the HP Ospo, you know, et cetera, so I've had visibility to many different Ospo's in large companies, small companies and startup environments. So I decided to, you know, collect all of this kind of information and produce a report that focuses on, you know, what is an Ospo, how can it run, you know, what are the roles and responsibilities and some lessons learned across the different experiences that I've had. And what's really interesting is that report, for instance, was translated to Chinese and to Japanese as well. And I've built a workshop based on this that I've delivered the workshop last year in Shanghai and I'm actually targeting to deliver it at the open source summit in Spain later this year. And all of these is really a collection of experiences that will help others see the mistakes that we've made in some of the Ospo's and how we were able to recover and proceed and also highlight some of the success stories. So, and the idea of all of that is, you know, if you are coming new into that space and you wanna establish an Ospo, we want you to be the most effective person to establish an Ospo. We want you to look back at our experiences, learn from our mistakes and learn from our success stories and instead of taking you five years to build an effective operation and be very well known in the space, we want you to go down from like five to three or to two. So basically bringing all that knowledge and unpacking it, making it available to everybody. Brian, thank you so much for taking time out today. And of course, talk about the Python Foundation, talk about the AI data. And of course, your reports and writing, as usual, I would love to sit down and chat with you again soon. Thank you. Thank you very much. My pleasure.