diff --git a/committee-steering/meetings/019-2024-08-29.md b/committee-steering/meetings/019-2024-08-29.md new file mode 100644 index 00000000..3825faaf --- /dev/null +++ b/committee-steering/meetings/019-2024-08-29.md @@ -0,0 +1,462 @@ +# Akash Network - Steering Committee - Meeting #19 + +## Agenda +- Review of community feedback from Special Interest Groups (SIGs) and Working Groups. +- Discussion on Akash project boards (Product, Engineering, Community, and Marketing Roadmaps). +- GitHub discussions update. +- Update on Akash community and marketing initiatives. +- Product and engineering roadmap overview. +- Open floor for discussions and community input. + +## Meeting Details + +- Date: Thursday, August 29th, 2024 +- Time: 11:00 AM PT (Pacific Time) +- [Recording](https://7rl44hv4pys2lipofrsxd22xmnvysihwvhfn3lttrx6w47u67w5q.arweave.net/_FfOHrx-JaWh7ixlcetXY2uJIPapyt2uc439bn6e_bs) +- [Transcript](#transcript) + +## Committee + +- Anil Murty, Overclock Labs +- Artur Troian, Overclock Labs +- Greg Osuri, Overclock Labs +- Scott Caruthers, Overclock Labs + +## Participants +- Tyler Wright, Akash Core Team +- Amanda Keay, Head of Marketing +- Andrew Gnatyuk +- Anil Murty +- Antons Kurakins +- Artur Troian +- B S +- Carlos A. Vera (At0x.eth), DeAI World Team +- Cheng Wang +- Christophe Lillo +- Denis Lelic +- Deval Patel +- Dominikus Brian, Akash Insider +- Fenil Modi +- FireLies.ai Notetaker John +- Florin Dzeladini, Deep Square Team +- Jigar Patel +- John DiBernardi +- Joshua Bate, DeAI World Team +- Juan Bujanda +- Julius Haus +- Kaylem +- Luca Esposito, Deep Square Team +- Luca Esposito's Presentation +- Luna +- Lussi Petelius +- Maxime Beauchamp +- Maxime Cyr +- oo o +- Piyush +- Robert Del Rey +- Rodri R +- Scott Carruthers +- Scott Hewitson +- Trigs _ +- Tyler Wright +- Zach Horn +- Zeke E + +## Meeting Notes: + +### Community Feedback on SIGs and Working Groups +- Tyler Wright invited the community to provide any feedback on the SIGs and working groups structure and improvements. +- No major feedback was provided during the meeting, but Tyler encouraged the community to share any feedback through Discord or other channels. + +### Akash Project Boards +#### Community and Marketing Initiatives +- Amanda Keay provided updates on recent events: + - Akash was the lead sponsor for the Cosmo Verse event in Buenos Aires, where the team was able to connect with local developers and promote Akash as a provider. + - Upcoming major conferences include Korea Blockchain Week, Token 2049, Mainnet in New York, and Cosmoverse in Dubai. + - Akash is sponsoring and hosting various side events and summits at these conferences. +- Robert Del Rey and Benjamin provided updates on the Zealy program: + - The first sprint is ending tomorrow, and they will share the final participant and reward numbers. + - The second sprint is set to start on September 2nd. + - They are collecting the best submissions and will share them with the community. +- Discussion around improving communication to Akash insiders about upcoming events and opportunities to support. + +#### Product and Engineering Roadmap +- Scott Carruthers provided updates on the improvements to the provider build process: + - The improvements are currently in beta testing by core community members and provider partners. + - The team is working on documenting the changes and preparing for wider deployment. +- Zeke discussed the Snapshot Service project: + - The project is in the final stages of stress testing before going live. + - The goal is to provide hourly snapshots of the Akash mainnet, sandbox, and other environments to assist with testing and development. + - The scripts will be open-sourced so that providers can set up their own snapshot services. +- Discussion around the upcoming Cosmos SDK upgrade to 0.47 and plans for community testing. + - The team is forming a working group to coordinate testing efforts across the community. + +### GitHub Discussions +#### Deep Square Proposal +- Dominikus Brian and the Deep Square team (Luca Esposito, Florin Dzeladini) presented their proposal to integrate Deep Square's HPC-as-a-service solution with the Akash network. + - Deep Square's goal is to leverage Akash's decentralized cloud infrastructure to provide HPC resources for researchers and scientific workloads. + - The proposal outlined a three-phase integration plan, including Deep Square becoming an Akash provider. +- Anil Murty, Artur Troian, and others provided feedback and suggestions: + - Concerns about the technical feasibility of integrating two separate orchestration solutions (Deep Square's and Akash's). + - Questions about the demand driver and how this would specifically benefit the Akash network. + - Advice to first try running Deep Square as an Akash provider to better understand the integration before proposing further collaboration. +- The discussion will continue on the GitHub discussion thread, and the Deep Square team was encouraged to reach out to the Akash team for further support. + +#### DeAI World Proposal +- Carlos A. Vera and the DeAI World team (including Joshua Bate) presented their proposal to use Akash for deploying large language models (LLMs) and providing ownership and rewards to users. + - The goal is to enable communities to have ownership and control over the data and knowledge generated by interacting with LLMs. + - The team is in a heavy R&D phase, exploring different deployment and monetization models for these LLM services. +- Anil Murty and others asked for clarification on the value proposition and how this would drive adoption of the Akash network. + - The team explained the focus on user ownership, data privacy, and building a rewarded ecosystem around LLM interactions. + - They emphasized that while in an R&D phase, they have made significant progress and have interest from partners and potential users. +- The team will look into providing a demo and continue the discussion on the GitHub thread. + +### Other Topics +- Tyler Wright encouraged the community to get involved in the various SIGs and working groups, including the upcoming SIG Economics and Website/Documentation working group meetings. +- He mentioned that notes and recordings from the meeting will be made available Soon. + +### Action Items +- Dominikus Brian and Deep Square team to continue discussion on GitHub and explore potential integration with Akash. +- DeAI World team to provide a demo and continue discussion on GitHub. +- Community members to get involved in SIGs and working groups, especially upcoming meetings for SIG Economics and Website/Documentation. + +# **Transcript** + +Tyler Wright: All right, I hope everybody can hear me. All right, again, Welcome to the Akash Network Steering Committee Monthly meeting for It is August 29th. 2024, Thank you to all those that are here as we do every steering committee meeting. This is a culmination of all the sigs and working groups that happen on a weekly monthly by monthly basis. Again, there's a number of special interest groups and working groups that Feel free to look at the calendar, which will be attached. I'm sure, sometimes soon, if you want to get involved, + +Tyler Wright: Usually we have an agenda that we go over. + +Lussi Petelius: but usually, + +Tyler Wright: I will start sharing that again. We start off with a review, a discussion of the community feedback around six and working groups. Then we discuss the Akash project boards. these are public facing project boards both for the product and engineering roadmap as well as for the community and marketing roadmap. So again, if you're looking for what the roadmaps look like ways to get involved, again, those are great places to start as well as the 16th and working groups from there. We'll go over, any GitHub discussions. This is a great place to get involved again. There's a number of folks that get involved between these meetings and GitHub discussions. I know there's a couple of folks here that are put up discussions that are going to be talking about those discussion today. Again there's a criteria that we go through for discussions based upon again feedback from the community length. + +Tyler Wright: Taylor Brown discussed him. And then if those participants are here during this call, we try to focus those discussions to again the focus around being around the benefit of the cost network. Again, this is a community open source call, so I will be making sure that focus talk about specifically benefits the caution at work during the discussion portion. But again, we'll talk about that later. If anybody has any other topics, they want to discuss with the steering committee or with the general community. Again, feel free to drop those in the chat and then we can open it up to those four questions at the end. And then we can look ahead to see if there's anything that anybody wants to talk about related to six working groups. As we move into September, which I know is a very, very busy month across the Akash network. Cool. + +Tyler Wright: To kick us off. I just want to see if there's anything from the community in terms of feedback conflicts. Anybody wants to discuss at this time. Feel free to raise your hand. + +Tyler Wright: if anybody again sees any room for improvements across the CIGS Working Group structure, or anything across the Akash open source framework, please feel free to drop a message and discord or, send others a message. If you feel more comfortable, + +Tyler Wright: All Next up on the agenda is the Project Board. Again, there's a community and marketing initiatives Project board as well as product and engineering roadmap, I won't go into a great deal of detail. And I won't ask folks from the Community and core team to go in a great deal of detail, especially around the product and engineering roadmap. Being that there was a SIG clients by monthly meeting where the console team gave really detailed updates. Are all the work that they're doing, but again, I'll make sure that we cover anything with the community or what's to discuss. + +Tyler Wright: I'm going to start off with the Community and Marketing Initiatives Project Board and then I'll jump into the product engineering robot following that I did want to see. Again, there's a number of initiatives going on, some things that I will call out There is a working group for the cost website and documentation that happens on a weekly basis for those that want to get involved in helping out with the open source Akash network website as well as documentation. There's a number of ways for you to get involved, please feel pretty, join a discord or join a meeting, introduce yourself. And then there's a number of issues, as we look to again on the document to our documentation side, continue to increase the user experience and improve the user experience for those that are using they caution network on deployment, side and provider side, as well as just general documentation on how to use. And then again, always looking for folks that + +Tyler Wright: Work Looking to Make Website Improvements. I know there's been a number of websites improvements that have happened as of recently, Denis. I just want to see if there's anything that you want to call out on the website front. Before we move on to some Q3 events prep that is ongoing. + + +### 00:05:00 + +Tyler Wright: Denis, if you're talking, you may be on mute if not. I'm happy to again. We can take these offline and talk about some of the changes. I know there's been some work that's been done to the ecosystem page. Again, if anybody here is working on a project that they want to have feature on ecosystem page, feel free to make a PR around the work that you're doing referencing, how you're using their caution network. And again, they'll be opportunities for you to add your organization, your project to that page. + +Tyler Wright: I know there's also been some other work that's been ongoing. So again if anybody wants to get involved with the working group around their cost website, talk about improvements that cost websites, get involved contributing to again the website, feel free to get involved in discord join the next upcoming meeting. + +Tyler Wright: Amanda. If you're able to talk, I know there's been a lot of events going on including cosmic verse that just happened recently in buenosaurus, and a number of events that are being cut for that. We talked about the community so would love for you to talk a little bit to the community about what's going on. And what's up coming? + +Amanda Keay: Zach from head of Marketing and myself went down to Buenos. Aires Akash was the lead sponsor for the Cosmo versus Buenos Aires Latin tour. We were able to speak and have a booth and give out a bunch of merchandise and connect with a lot of local developers down there, really looking to get some more providers down in the last time community. So that was really great for us and then looking forward, I'm pretty sure we have six major conferences in the next six weeks. Most coming up is going to be Korea Blockchain week, which is next week. The middle of September. We have token 2049, which a bunch of our team is going to be at and we're sponsoring that as well. + +Amanda Keay: And then moving to Massari Mainnet in New York City. The the very end of September into the beginning of October a cautious sponsoring that, as well as hosting a series of side events and depend summits as well. And then moving into October, our team will be out in Salt Lake City for permissionless and then + +Amanda Keay: I guess over to Dubai for Cosmo verse the main conference. This year, a little into November we're still deciding on a couple of events we are potentially heading to DEFCON in Thailand and then doing a series of AI conferences, Ray Summit as well as near IPS into the Q4. So Our team is going to be traveling most of this end of the year but really looking forward to it. We've been having a lot of really great feedback from people that are looking on and saying that Akash is everywhere when they go to a conference which is great to hear. Greg has had Multiple speaking engagements at all these conferences. So, really happy to be pushing pedal to the metal with the events. + +Amanda Keay: Does anyone have any questions about anything upcoming? I do see a question from Andrew. So Luna. Breakpoint is going to be in Singapore as well. I believe it's a couple days after token. I think we are able to attend + +Tyler Wright: Thank you. Amanda again, a number of discussions happen, around events during SIG community as well as the working group for the events 2024 Discord Channel, much appreciated, Amanda and all those that help put those events together, support on boots on the grounds. Again, there are a number of opportunities for those that are akash insiders and have availability to supported these events again covered by the events proposal that went on chain in the past, a number of months back again if anybody wants to look to get involved in the cautious inside program that isn't please feel free to look at that cost network website and believe there's a typeform for you to apply. And then again, for insiders, you all know. There's a number of opportunities for you to support some of these events and + + +### 00:10:00 + +Tyler Wright: Support goes a long way with all the booths kind of been etc. All the programming that Akash does so much appreciate everybody. + +Tyler Wright: Excellent. Is everyone have any questions on Q3 Events Before I move on? + +Tyler Wright: I know there's a number of other initiatives going on, again, I believe, since our last meeting, the Zeely proposal has passed, and I know that there's been a number of daily missions that I've been Shout out to Robert and Ben Benji for all the work that they do around. The developer onboarding program using Zealia as a tool Roberts real quick, I just want to see if there's anything that you want to touch on in terms of Zealy + +Tyler Wright: Go ahead, Robert. + +Robert Del Rey: But yeah, on really John, it's been very active guys. Just yesterday I was looking at the analytics and getting some numbers out. Benji and I are putting together one month report where we want to show you how we did on this first month. so yeah, we started getting the information, the screen number one, it's gonna end tomorrow. So once the screen ends we're gonna have the final number of participants winners and the total amount of akat rewards for The second screen is set to get started on September the 2nd, which is next Monday. + +Robert Del Rey: So yeah, that's the most important stuff that we're doing on Siri right now, we release some missions by mid-month. And overall we're very impressed by the participation and also the quality of the submissions. let me say real quickly. We also have a document where Benji and I are adding the best of the best submissions and hopefully we share that with you and more things that we're cooking in the kitchen. so yeah, that's what I wanted to say John on and thank you for all the guys who have been getting involved. A special shout out to B s to Rodri and of course when you and I don't know… + +Tyler Wright: No,… + +Robert Del Rey: if you want me to go over YouTube. + +Tyler Wright: don't worry about that. + +Robert Del Rey: Now that happened + +Tyler Wright: We could talk about that during the next design meeting, that's just some more high level effort. that's good. + +Robert Del Rey: Right. + +Tyler Wright: I want to focus on some more high level efforts and we can talk about that in a future meeting when we have some more deliverables. I know that there's a fair amount of work that's been going on around the chat API again. + +Robert Del Rey: All right, all right. + +Tyler Wright: There's a couple of discord channels for those that want to leverage that, as well as a number of other efforts. I just want to see if there's anything else on the Community and Marketing Initiatives board that anybody had any questions about or anybody wanted to discuss in more detail before we move on to the Park and Engineering World map. + +Tyler Wright: Cool. + +Tyler Wright: Go ahead, Robert. + +Tyler Wright: I think that's a great idea. We'll talk about it, certainly at the next entire office hours or on the insiders channel if you want to discuss it more Robert. But I really appreciate that feedback. + +Tyler Wright: Anything else on the community and marketing side before we jump into some product and engineering issues. + + +### 00:15:00 + +Tyler Wright: Go ahead, Carlos. + +Tyler Wright: yeah, we're gonna have some time during discussions to go over the discussion that you all put forth and go to + +Tyler Wright: Yeah, no problem. That should be happening in the next. Couple of minutes following this product, and engineering course. + +Tyler Wright: All right a couple of things to discuss again. There's been an ongoing process led by Scott Carruthers from the Overclock Core engineering team around improvements to provider build process. These improvements again are in beta form and been tested by core members of the community. The various different concept providers again, there's a product that's a plan that's been talked about in sick, providers in terms of next steps. If anybody wants access to those provider build, kind of alpha documentation because it's not an official documentation yet, as it can continue to go through testing and some different applications and some console and other tooling then again, feel free to reach out and seek providers and providers channel and get access to the alpha documentation. + +Tyler Wright: One other efforts that we talked about in the past. I think we'll talk about a little bit further during the discussions. It's around the work on the Mystic Lab team has been doing with the Metamask and Cosmo kit. I know that they've been going back and forth with members of the console team. And that work is again, there's videos that have been created and it's been tested. I think there's a plan in place for merging that but there's no other deliverable that at this time is needed from the Mystic Labs team. Again, this is an effort that they talked about to previous steering, Committee's report to proposal on chain and have again since completed. The work have been very responsive to work closely with the console team. Again, we'll talk a little bit about some of that work, a little bit later during the discussions portion, I'm at the beginning of the discussion portion, but just want to call out that work and it's still actively and testing, but has been tested by the console + +Tyler Wright: Even those again, a plan in place for implementation down the line. + +Tyler Wright: Again, there's a number of other efforts that have been discussed in great detail. During various meetings, including Sid Clines, including a sick chain, as it relates to the upgrade for the causes SDK to version, 0.4.7 or scheduled, for 0.47. Im I think we'll talk a little bit about that during discussions as well. But from a technical standpoint, I just want to see if there's anybody here. That wanted to discuss anything on the product and engineering side that maybe you want to talk about in more detail or just highlight here during the call. + +Tyler Wright: Feel free Zeke to talk a little bit about, I'm sap South. I know we talked about this during SIG chain and some other meetings. This is a project that Zeke from the Community Validator insider. Where's a number of different hats has been working on to provide again with snapshot service to the community that grab snapshots on an hourly basis. I know we have some other snapshots services out there that capture snapshots for the crash network Mainnet, Sandbox another environments on a less regular basis. But again, for various testing purposes and just as a tool for the community, in general, to have again, Zeke from the community work, from a spec from the core engineering team. And again, it's in this final stages before going live and being available Widely to the community Zeke, is there anything that you want to talk about more specifically there? + + +### 00:20:00 + +Tyler Wright: Thank you very much again for those that want to get involved, there is a group forming for those that will be doing various testing for the cost of DK upgrade. If you are technical, please feel free to reach out to me or just put a message in chat. And again, you can be added to that group to do some initial testing. The part of that testing Again, we went into great detail about this. During We want the group wants to test all aspects of the network. So we'll pull together a number of folks within the community to help with that testing before we do some more specific stress testing with other partners after that. + +Tyler Wright: And again, please look out for a number of changes that are always ongoing around console. There's an ecosystem, console discord channel. There's a console repo where, again, the console team that the great job of following up around issues constantly making improvements to the product and sharing those improvements. So again, if anybody has any questions or this client by monthly meeting or feel free to drop a question, comment, etc inside of the ecosystem console channel. + +Tyler Wright: if you want to contribute, I believe there's some items that get tagged, it's good. First issue for those that want to get more involved in again contributing to the Akash network again, please. Feel free to look inside console issues and see about Any good first issues and please get involved in the chat before moving forward with those issues. So you can get some feedback from the team. + +Tyler Wright: If anybody has, any other questions around the product and engineering roadmap. Again, feel free to Any questions as comments inside of any of these specific issues, this is a public facing document. So again, as members of the core team and members of the community, continue to develop and build their updates, that are made available inside each of these issues to consistently, track the progress. + +Tyler Wright: Anyone have any questions or anything else? They want to discuss on product and answering engineering roadmap before we move on? + +Tyler Wright: All right, next up is discussions. Again, during this portion of the agenda, we look try to have folks, talk about their project. I think that number a couple of folks that's asked to speak just for just ease of operation, I'll hand it over. I know we talked a little bit about the work that Mystic Labs has done so we'll talk about we'll discuss some of their work a little bit pass over at Huey first. And then after that I'll hand over the mic to the + +Tyler Wright: Domi and his team to talk a little bit about on the work that they've been doing following that. Then I'll handle over to Dsi team to talk about their discussion. And then finally, if there's time, we'll talk about the proposal that thing just put on. So again, that'll be the order of operations. If anybody has any other discussions that they want to talk about more specifically, please feel free to drop a message inside the chat again, because of the sake of time, and because of this audience, who is very much focused on their growth of the cost network. I beg that the discussions are focused on integrations, improving their caution network, and those kinds of subjects if they start to be our off and to a little bit more details and a little bit too focus. Then I'm just gonna raise my hand, which will + +Tyler Wright: Politely give you 30 seconds to kind of wrap up that discussion. Again we want to have these be as high level and focus as possible. If anybody wants to get into the details that's what these actual discussions are for. But I do want to give again teams that are leading to discussions and have been involved for some time to have some time in front of the steering committee to talk about these things more specifically. With that said, I do want to hand it over to Scott as a follow-up with some of the work that the mythic labs teams has been doing. + + +### 00:25:00 + +Tyler Wright: Anyone have any questions for Scott the team behind Mystic Lab, obviously a much, appreciate all the work that they've been doing. again you can see the original proposal, they put out there as well as the follow-up work. They've been doing kind of track all the work that they've been doing much appreciated again, they're availability the response time and just again, all the work, but does anyone have any questions on that, follow-up proposal? Before we move on, + +Tyler Wright: Thank you, Mr. Gladstone All right. Next up, I want to hand it over. I know this is a proposal that doming from again, their caution ciders has been talking about and for some months at this point. Again, I want to hand the floor, over to Domi and the team to talk about the proposed,… + +Dominikus Brian: Okay, thanks Tyler. + +Tyler Wright: the proposal at the highest level on the benefits to the cost network,… + +Dominikus Brian: So yeah, by the way, can we share a little bit of some quick,… + +Tyler Wright: and then we'll open it up to any comments or… + +Dominikus Brian: presentation? So you get the idea better? + +Tyler Wright: questions real quick. + +Tyler Wright: As long as it's quick, I know that we've had these presentation that are made available in the discussion that hopefully everybody's been reading for some time. + +Dominikus Brian: Yeah, it's great. + +Tyler Wright: But yeah, I'm feel free. + +Dominikus Brian: Okay, look how we'll put it up. So for now, I'll first in introduce a little bit, so the motivations behind it is to really have HPC client that. Can be served on top of So instead of DOCKER specific or the whole V instance specific, it could be taskbase. So sometimes people doesn't need the whole GPU or maybe they want to fit several jobs on the GPU. So this is the kind of technology deep squares working. That's why I'm looking for how we can possibly integrate this kind of technology stack into sh. yeah so today joining with us there is a deep squared team. Luca Florida and Lillo they will have a quick pitch on this and then will taking The whole idea how this can be beneficial for our cast and so on. Yeah look up. Let's take + +Luca Esposito: Thanks Hi Taylor. Very nice to meet you. And thanks to the old Akash community for all singers. So our vision to integrate, deep square solution into your cache network is to really capturing the growing C market and democratization of access to HPC for AI acceleration. Our mission. Really is to enable the Attach network with the integration of the Deep Square and HPC per minutes capabilities by simplifying and accelerating the user experience. Again, this is really connected with the growth Tellery we're talking about. And so, we kind of foreseen the project timeline dividing, the Phase One, where the deep square will become in a cash provider and start to design a framework integration of our solution for rent, HPC per minutes into the network, and of course, face suits, kind of integrates the framework into the + +Luca Esposito: Cash console and make it available to the cash providers and Phase three, which would be the deep square plus the cash blockchain functional disintegration. So for this, of course, we are kind of requesting for each phase, a 50K budget for developments and incentives. And we kind of really believe that this integration would be extremely helpful to your casting. Also, considering the grades market request for HPC next few years, as projected, by a line market research, which is projected to be around 17 a billion SPCA service market cap. So, we also seen quite a spike in the month of April, John's network with the user deploying, literally thousands of CPU using instances leads to do some work, that's really would require, which Deep Square can help to kind of capture these. + + +### 00:30:00 + +Luca Esposito: Then distributed through. + +Luca Esposito: The HBC infrastructure. So the motivation objectives are kind of clear but we want to bring to the network. We are a number of features around parallelization and improve the provider HPC building criteria, provide multiple jobs spans over many CPUs and hardware and resources. Verification Of course, this would need to kind of test those capabilities in the face one and of the projects. and of course kind of test and analyze both sides products architecture. Just to thank you very quickly through what we provide is deep square. We have a multi-layer proposal where the applications, they connect to our platform through APIs and Web Portal. And then it goes down to the layers of validators, supervisor made the schedule. This is where all the workflow and the blockchain capabilities. They kind of kick in and they all through connected to our interoperability layer. + +Luca Esposito: Called Cluster Factory, which is an open source technology, we have developed and used by providers and clusters and clouds. And this is where we see a cash really kind of playing a role into this and becoming, Pretended Green to integrates our solution into the network just to literally apply over on some of the clients user interface with a Web portal. we're developers can easily and safely get that and access to launch workflows in a very kind of easy mother Again through to a Web portal and also pay on a permanent basis, whatever infrastructure they use. We also have use cases for render streaming which they run on our infrastructure with a video on YouTube which kind of show can do. Same thing is for text to image, AI interference which is another use case running on our + +Luca Esposito: ure. We provide deep squared SDK package. + +Dominikus Brian: Station flow. + +Tyler Wright: Perfect. + +Luca Esposito: All right, I thought I had seven minutes but nevertheless. Okay. + +Tyler Wright: Excellent, no much appreciate. If you want to share that presentation again inside of the Github discussion, that would be great. I saw Archers hand up first then Anil's and then Zeke's Archer you go first. + +Artur Troian: things that are all typically a couple questions, number one is, so I have actually mentioned Presentation in the just to let you know that the GPU sharing at this stage is not supported on a crash. different number of reasons. So I don't know how + +Tyler Wright: And then Anil have you follow up I know Anil dropped some specific notes inside a discussion so much appreciate that but Anil, I'll go ahead. + +Anil Murty: He told me and ca. So I guess, first off, thanks so much for the writing of the proposal and for this presentation as well. It's nice to look through your solution and stuff. I should have probably done this sooner so apologies for doing this yesterday but I spent a bunch of time It probably read through it three times to try to understand what was being said. And then I also looked at your documentation in any other resources, I could find on Deep Square itself. And I kind of have a lot of questions, but a bucket them into three areas. So the first is, when we typically look at integrations or we look at, proposals for building something akash, Our primary objective is to try to drive demand for a cash. That's kind of the number one, right? + +Anil Murty: and in the past, most of the projects that have gotten funded by the community through, the proposals in discussions here, The primary focused on building some sort of higher layer application, that doesn't mess with the underlying protocol, or with the way orchestration is done, right? So my first question basically is in that area, which is when I look at Deep Square today with the minimal understanding, I have a deep square. What I understand is that you basically built a network that is in many ways, similar to Akash except your focused on a much narrower subset of users. Your specifically focused on + + +### 00:35:00 + +Anil Murty: research institutions that use HPC which was the primary driver of accelerated compute until about two or three years ago when Tag GPT came, right? So before charging, we came along for everybody's context, The primary driver of GPU compute usage in the world, was these big national labs, like Sandia National Lab and another places that would consume a huge amount of CPUs running tasks across, thousands of clusters CPUs. so given that you build that solution, you have your own orchestration method that uses something called And in many ways Norm is a technically, competitive thoughts or an alternate solution of sorts to Kubernetes or really and + +Anil Murty: Most people that are running quote-unquote production commercial inference applications for AI today. They typically use Rey on some sort of underlying orchestrator or, straight up on VMs and that's the path that they take. So, the broad questions I have are two broad questions. I have one that is number one, This is very different from other integrations that you have considered in the past which are not messing with the underlying protocol. This is almost like trying to take two different orchestration solutions and try to merge them together, which I think is a bad idea. + +Anil Murty: Number two is How is this going to drive demand? The only way this is going to drive Demand is if you already have a bunch of researchers that are using deep squares solution today that are on or a bunch of researchers that have said that they'll use deep square solution today but they cannot use it because you don't have enough GPUs because I kind of extracted from the other thing I started from the right up is that you have a cluster of 12 GPUs today with the primary 30 90s and obviously, a lot of these people that are doing these massive simulations. Need a lot more GPUs in that, so that would make sense. And if that's the demand driver, it makes sense. But then the question is, you actually have a bunch of research institutions that are signed up inside. Hey, we're going to use deep square if you have on your GPUs. And so you're saying, okay, there's a lot of technical challenges to making it work but Is the business case there. That's kind of number two. and the number three part of this is + +Anil Murty: in the proposal. The very first phase talks about deep square potentially becoming a provider on our cash network, + +Dominikus Brian: Yeah, that's Luca. + +Tyler Wright: Perfect. And… + +Christophe Lillo: That maybe I could. Sorry. + +Tyler Wright: But you can go first and then Luca will follow Zeke. your question if you could drop it in the chat, that would be great. I want to spend maybe three or four more minutes on this discussion before we move on to some other folks that I've been waiting. So I'll hand over to Christophe. + +Christophe Lillo: Yes, thank you. Just to bounce back about Kubernetes and slam. we are already running Kubernetes on top of Sloan. the strong cluster is the completely deployed on humanities and why we use this scheduler basically because we are really focused on parallelization and let's say true HPC jobs. Where, for instance, for milky nodes with multiple GPUs training and I would say also just to give a little bit of information, it's also a container as a service but I would say an HPC Companies service and we have compatibility with Docker and every basically container runtime who are using up and route from NVIDIA. But yeah, it's containers in + +Florin Dzeladini: Yeah, and maybe if I can add Anil, thank you so much for looking into deep square and for asking the questions. So I think the reason why we want to become a provider is that we want to understand what it means to be a provider because we think that where we add a lot of value is on this cluster factory stack, the way we integrate Kubernetes with Slurm. And the way we allow people to do this multinode with very fast, interconnect type of jobs and how this could then benefit to So by becoming a provider, then we have to go into the meat, And understand how it providers sets up, the Kubernetes and everything. And I think the goal for that is also to get to know. + + +### 00:40:00 + +Lussi Petelius: I think. + +Florin Dzeladini: Also what's the stack and how all this could work together with cluster pattern? and then,… + +Artur Troian: Yeah, I'd like to intervene a little bit here so important for just interruption. + +Florin Dzeladini: on the side of the + +Florin Dzeladini: Yeah, no worries. + +Artur Troian: Just like probably my summary here. Make sure that you guys going correct way a thing you just mentioned that you want to try, what is been across from what I got provider. So you don't need the governance proposals for funding or incentives here, you can just spin up provider, quite easily, hook, it up to sandbox and all the information is public and you can even just spin up your blockchain locally and spin up. The provider Kimberly is locally and just see how everything works. And we'll be honest, our development environment has been Documented quite well, I believe and All the things that allow you to just in a provider are so easy. Thanks 10 minutes to set it up. So I quite sort of probably curious how + +Artur Troian: Across community will be benefiting from this things at this point. If you guys don't let me just say clear don't have full understanding how God's provider works. So I would say The next,… + +Florin Dzeladini: Yeah. + +Artur Troian: let's probably first, give it a try from your side just to make sure that you understand how a car is centralized Cloud works from our It's a technical perspective and then see how things that you try to actually see the teacher that architecture. And then I just don't believe it's going to have quite a lot of changes to the proposal that you might + +Florin Dzeladini: And I understand that. Yeah, makes a lot of sense. But what I'm saying here is that basically is how to kick the ball because at the end, what I see is that you have this manifest files and we have this workflow file system that allows people to spin up these jobs on slurm. and this is very similar. I mean it's a yaml file and everything. So when we were looking at this with Lillo and looking at the stack of akash, it was directly coming to our mind. but there could be a v3 of this manifest. There could be a new version of this manifest which could include those features and aloe and user that is using a cash console to then be able to spin up jobs. That can take advantage of this slur that can take advantage of this internode communication. + +Dominikus Brian: Okay, thank you Florin. + +Tyler Wright: Thank you. + +Dominikus Brian: So yeah, look at you. Maybe can answer quick on the demand side and then we move on. + +Tyler Wright: Of outlet Anil. + +Florin Dzeladini: And maybe in 10 seconds else. + +Tyler Wright: Go, I'll go. Yep. No please please Florin go. + +Florin Dzeladini: Sorry. Yeah. I just wanted to just add something at 10 seconds, it's just that you see, we have some people that are interested and do. thank you so much and the way Deep Square. Basically, we're trying to find providers for deep square, Because we need more than just the GPUs we have in this proof of concept immersion cooling system. So that's where basically, with dummy we're discussing and we were like, but there is a lot of GPUs that are unused at in Akash network itself. So if we could also on the other side, let the providers understand what is close to factory and be able to provide GPUs to potential scientists. Then this could be also very beneficial to the Akash network and then how those things then take that. And how the two things would merge then. This is very, very early, right to understand how this could work. But what we can do easily is aloe provide us to join the grid and allow the console to be able to spin up HPC jobs. And I think with that we would be able to enter in a new market with that Akash is not nothing in + + +### 00:45:00 + +Florin Dzeladini: Now. + +Tyler Wright: Perfect. Thank you il. let you get the last word and then we'll want the discussion around do sideworld. + +Tyler Wright: I do want to move on to go ahead or to real quick and then we're going to move on to the D cyclist discussion. + +Artur Troian: Yeah. + +Artur Troian: Yeah, I'll just gonna back What Anil said on that is. We have quite, API the dogs. You can just create deployments. So that's number one. Number two, is just remember, one thing, probably related to finding that everything related to funds that has been requested all the integrations are called basically open source. So just total reminder, I don't know what your + +Tyler Wright: I know Just dropped in links to the API as well as the cost JS which are two ways for again you to integrate with the cash. I would love to again, This is lively discussion. I want to continue this discussion in Thai Github again. It'd be great. if you have any questions, feel free to reach out to me. We can set up a channel if you all need to get some more support. But again, much appreciate the discussion much, appreciate you all being here and talking a little bit about live. I do want to leave again some time we do have a number of other discussions to get to and we only have about nine minutes left. I want to be conscious about everyone's time because it's very, very for some people very early for others. So again, I do say a hand up. If you can leave your question in the chat, that would be great or leave it on the specific discussion. I do want to hand it over to the Desai World team. I hope I'm saying that right to talk about highest level. + +Tyler Wright: On their proposal. Again, the benefits, the Akash network, and then we'll open up to any follow-up questions. And then following that, those couple other discussions,… + +Dominikus Brian: Jobs. + +Tyler Wright: I want to see if there's any open discussions from there that will probably have to take offline and then let's go hand over to Deeside World team to talk a little bit about the discussion that they put forth a little while back. + +Carlos A. Vera (At0x.eth): Thanks hey everyone, and thanks for the nice comments on the discussion so far, really enjoyed it. So on a high level, what we are trying to do is do a lot of product research in novel ways to deploy on models. Namely open source stuff like what you're seeing with 1131 up for communities to have ownership over in order for them to create their own sort of persistent memory clusters, right? This has let us from. + +Dominikus Brian: it's + +Carlos A. Vera (At0x.eth): These are world team into their cash solution as the best current. + +Carlos A. Vera (At0x.eth): way to deliver that to our partners and we are currently in a very heavy R&D phase where we're trying different deployments and different solutions all hosted in a cache. so just to lead the conversation for today, what I won't really want to do scope it out because our main issue right now is mainly tokens for inference. So if we can have enough tokens to try out New mechanisms and different looking for oasis product. Will eventually be able to offers so serve as a service for people where these would include an employer's group or you would launch their own provider because that enables them some degree of privacy. I did want to ask a question about the stack and to what level the provider has access to what's going on on the + + +### 00:50:00 + +Carlos A. Vera (At0x.eth): The cluster as in even a friend is for medical practitioner for it to be his LM for his consult. Can the providers snoop into what's being said and… + +Dominikus Brian: just, + +Carlos A. Vera (At0x.eth): Mainly what we want to do is to try these things out and practice. We have Both on channel of chain groups of people both scientists academics and people who have this kind of needs that we want to explore with and then come back and show you guys success stories that can drive the cash adoption. From a more users, that consumer site rather than large institutions, is an inference So yeah, Joshua's here too. So I don't know if he wants to say something but earn that open for comments. + +Tyler Wright: Perfect. I just want to see if anybody had any questions. I see two hands up I think Anil was up first and then a Artur will fall Anil. Please go ahead. + +Anil Murty: So I was reading through it and I meant to post comments later. I didn't get around to doing that, but I'm just having a really hard time understanding What the value proposition here is and then why would somebody use this versus using? the hundred other Llm hosting services that are out there in the world right now. + +Carlos A. Vera (At0x.eth): Right. So, specifically the value proposition, we want to offer users is to have ownership over their data, and they're the production of the knowledge that when they interact with these agents, they're producing useful information that currently goes into a black box and people don't know what happens with that data, after it reaches People's Server. So you're being hosted like everyone's saying that they're not training their models on users data, but they all are. So, the main value proposition that we want to offer as a protocol ourselves is to give these ownership certificates against claims that are generated between users and LMS. And also give these ranking of usefulness, which you can only be created on an open sort of market where we can create these sort of just in time, vector stores, + +Carlos A. Vera (At0x.eth): Were inference for different use cases. So the main idea that we're trying to sort of Get communities on board with is that they start creating knowledge about the topic that they really like, but that knowledge, a part of it can be used for other communities, but in a rewarded manner. So over time, + +Carlos A. Vera (At0x.eth): These communities create a specific profiles and personas. That can be fetched by the protocol at large to create a more nuanced response for users because a lot of the research currently is saying that you don't necessarily need larger models, but better context and maybe multiple step inferences. And by having wholesale computing available and you're not being charged for prompt. For example, you can do a lot different things and also do back and prompting. So to speak where you're taking snapshots of your data, you're creating summaries you're creating these real time, knowledge of what's happening, and that can be used as value for your community over time. And mainly, that's the other thing we're trying to use is a coordination layer. So by having these outputs sort of outing, the open other networks, kind of touch to them and create more meaningful knowledge basis, like composing them. + + +### 00:55:00 + +Carlos A. Vera (At0x.eth): How that answer? It's partially. + +Tyler Wright: If you need to drop off to go to some other obligation again, feel free to in one minute, but are to go ahead, go fast. + +Artur Troian: Yeah sort of question you mentioned initially some Akt tokens to run the deployments. + +Carlos A. Vera (At0x.eth): Yeah. I mean, mainly we're trying to have these employments long-term. For example, we have several events that want to have their own agents during the events and we've been host, we've been for ourselves for these things, but we would like to have some R&D budget, basically that we can use to try these new mechanisms out, and sort of get to the point where it's a product that people can just deploy by themselves and sort of Babe $50 to have five days of LM with all these services,… + +Tyler Wright: Okay. + +Carlos A. Vera (At0x.eth): and it's powered by a cash. but to get there, we need to figure out what the right sauce is so, sorry, Tyler, go ahead. + +Tyler Wright: No, no you're fine. I would just ask that maybe again for specially, some of these discussions that I'm projects that maybe a little bit earlier on we can talk about some of these things in sync providers or again, SIG clients and discord channel. We could talk offline about, helping with some initial R&D, if that's the next phase. But again, I do want to leave time for a couple other discussions. I would ask that anybody wants to get more involved in these discussions. Again, participate in Github. That's what it's there for, it's open. Anybody can create a github account on if they don't already have one and contribute, I'll go ahead, Joshua, get the last word, and then we'll continue to move on. + +Joshua Bate: Thanks. We're in a time crunch. I just want to reiterate that this whilst we're talking about R&D, we're talking about this in a sense of We want to optimize a product. So that it's scalable for long-term future adoption rate. That doesn't necessarily mean that it's not already something useful. We've been working on this for quite a while over a year. Now we've got a lot of developers at working on this and we've been providing a really great vision but actually backing it up with an early foundation of a product that's usable. We're working very closely with a lot of partners globally. myself personally. I basically have just been doing BD for my whole life in crypto and basically, no most communities in the decentralized space and everybody wants to use this thing. And so which it does now on smaller scales when it works out large, there's gonna be a lot of demand for these instances, which if we're powered by a cash, is a lot of demand for a cash VL and to be spun up. When Carlos says R&D, we're being humble, right? we actually do have quite a decent amount of + +Joshua Bate: Progress behind us and this is what sort of I want to reiterate that. Because I understand it seems like we sort of should be an early stage discussions, maybe chatting a discord with mods and stuff. But I do want to reiterate, this is actually quite an advanced build that we're working on and we do have funds of our own that we can, put into this. But we're a very early funding product and we're trying to basically optimize our funding workflow and we thought that this would be a nice way for us to bootstrap some of the immediate requirements and develop a nice partnership with what we think is the best provider of this tech in the space. And so I just want to make that clear that we are actually quite an advanced product at this point. And I hope that comes across + +Tyler Wright: I'd much appreciate that. I think a nail put a message in the chat around a demo. I think if you had a demo that would be great to add to the discussion get some feedback from the community in sync clients. I can also add you all to the SIG clients call that I think is coming up in maybe not sometime very soon. And then again, You have my information Tyler core team. You can reach out to me and discord to talk about something. Again, a demo would be a great. + +Tyler Wright: again I'd much appreciate everyone's participation today. I know that everybody has plenty that they would like to talk about relative through specific project. I did want to give folks here that had some discussions up for some time, they're a bit of the ability to talk in front of the steering committee as well as live to the community. Again, a lot of people listen to these recordings later, I'll make it available on very shortly, florin. I know you might have another comment if you can drop it in the chat because I do want to get to one more chat and one more discussion and then very quickly and then to open discussion from there. + +Tyler Wright: again, there's a number of discussions, much appreciate everyone taking the time today. If you have time and general life, please get involved in discussion between meetings so that we can get the necessary feedback prior to the steering committee. And then the groups putting together a discussions, can answer very specific questions, provide deliverables if needed etc, to manage everyone's time and efforts. But again much appreciate those that presented today. + + +### 01:00:00 + +Tyler Wright: I know that there's been a new discussion, put up. We talked about the Cosmos upgrade a little bit, I know it just went up. So again still I want to get people time to read it to go through it a little bit more detail. And then get involved in the comments. I'm sure you'll be talking a little bit more during future events Cheng. I just want to see if there's anything that you want to highlight real quick. I know we're way beyond time at this point but just wanted to make sure you had some time to talk about the discussion. + +Tyler Wright: Again, please get involved in discussion. Cheng just talked about as well as all the discussion. I'm sure they'll be some additional feedback that comes from it. I will drop in the recording and the notes shout out to George for all the support. He provides from the community they caution center with helping with the notes, Those notes will be available with the next couple of days. I'll drop them in discussions for anybody. That's involved in the discussion to reference the steering committee Here today. before we jump off, I just want to see if there's any other open discussions that anybody wants to talk about real quick, we'll probably have to take them offline but real quick. This is anything that anybody wants to discuss around. Akasha hasn't been discussed. + +Tyler Wright: All as we look ahead again, there's a working group around that cost website and docs coming up soon as well as SIG economics. I think happening next week. again for those that are feel free to get involved in some of those meetings. Send a representative from your team to some of those meetings. Whether it's think providers to learn more about provider, infrastructive there's a number of different levels of providers that join that call. And so again there's no better group to talk about providers and understand what's going on underneath. Then that sig there's a clients by monthly on. There's a chain monthly, there's a number of six, they're talking about everything from the process SDK upgrade to console improvements, the provider and everything in between. If you have some non-technical abilities, again, there's very CIGS and working groups to get involved with there to Continue to push on awareness around the Akash network and beyond. So there's just so many ways to get involved. + +Tyler Wright: Again, much, appreciate everyone, taking some extra time to continue his call, running a little around a little bit long, thank you to all those that presented during discussion Today. please again, I'll drop in the notes. If you have any other additional questions, thoughts, concerns feel free to drop those inside that you're respective discussions and we'll continue to work together in between meetings and discord and elsewhere. But again much appreciate this community all that's going on. Please let me know if you need anything more specifically happen to be a resource for you. Hope everyone has a great rest of your day. Thank you to steering committee. Thank you to everybody in the community. Thank you all. + +Dominikus Brian: Thank you. All. + +Robert Del Rey: Thank you all, stay safe. + +Cheng Wang: Thanks everyone. Take care. + +Christophe Lillo: Thank you. + +Cheng Wang: Next time. + +Rodri R: To you. + + +### Meeting ended after 01:07:16 👋 + +*This editable transcript was computer generated and might contain errors. People can also change the text after it was created.*