In this Ecosystem Aces Podcast episode, Chip Rodgers, CMO WorkSpan is joined by Manish Harsh
MLOps Partner Ecosystem & Tech Integrations, Nvidia.
Manish Harsh brings over 20 years of experience in advanced technology, including cutting-edge areas such as 5G, edge computing, autonomous vehicles, and machine learning operations. He is currently building an AI and ML partner ecosystem, community and industry vertical alignment specific to use cases of scale and production at Nvidia.
Topics covered include:
- What is Nvidia’s current charter? - 2:23
- How do Nvidia's partners use Nvidia's platform for innovative technology and end-use cases? - 6:25
- How does Nvidia work with partners to understand their requirements? - 9:10
- Nvidia's vertical solutions in AI/ML and cybersecurity - 14:36
- Overview of Nvidia's partner programs and how they are tailored to specific partners? - 19:38
- How does Nvidia manage joint roadmapping and partner collaboration with product teams? - 22:08
- Advice for working with Partners - 27:27
Chip Rodgers 00:08
Hey, welcome back to another episode of Ecosystem Aces. I'm Chip Rodgers, CMO at WorkSpan and excited to have a conversation today with Manish Harsh. Manish welcome.
Manish Harsh 00:22
Thank you. Glad to be here. So good to see you again.
Chip Rodgers 00:27
We had a really interesting conversation in preparation for this. I'm excited to get into some really interesting partnering topics around some hot areas these days and that you're right in the middle of. Let me start with a quick introduction of Manish. Manish has been with Nvidia for about nine years now. And as I said, has been deep into a lot of really interesting areas starting 5G and edge computing and working with the developers on the developer platform. Then also working in autonomous vehicle topics as well and most recently in the machine learning operations, partner ecosystem.
Prior to Nvidia, you also were for many years in similar areas like Halosys, Celestial and Corbus and so have a long history and advanced technology and some really cool engineering. So Manish welcome
Manish Harsh 01:49
Thank you, Chip. When you went through all this LinkedIn, I felt like I have done quite a lot. Time flies. But thank you for the introduction and I'm glad to be here.
What is Nvidia’s current charter?
Chip Rodgers 02:06
Well, Manish just start with the things you're working on today. And tell us a little bit about what you and your team at Nvidia are up to and what's hot these days?
Manish Harsh 02:20
So my current charter is actually very simple. We make friends, we make true friends. And we are building partnerships. And as you know, we all realize business is happening in a very new way now with the ecosystem. This is Nvidia's philosophy, we rebuild ecosystems around almost every vertical. And today I'm sitting in a data center, which is a data center business unit, which is called Enterprise product group with an NVIDIA company led by Manuvir Das, who is a super boss. And my immediate boss is Scott Macklin. And we have a small team, which we all focus on AI platforms in the ML ops ecosystem. That's our charter.
It's a pretty large landscape, around 300 plus companies, New Age companies who are building software for different personas. We want to ensure that they have the right tools, right understanding ,right knowledge about how they can leverage Nvidia tech. That's step one, we want to understand they understand us well. We are a platform, their platform or our platform for ML engineers, DevOps data scientists, AI practitioners, data scientists now. So we cater to the persona and we understand the workloads they are dealing with.
Now, why I was telling the story is because we want to ensure we are building an ecosystem of these ISVs who serve all these personas. Our charter is specifically to qualify them, learn about them, and ensure that they understand our tech well. Once we get that milestone done, we will go to the next step. We will ensure that we give them the right GTM, go to market, business support, roadmap visibility, learn about their roadmaps and ensure that we both are growing together. It's like a coral reef kind of analogy, where we 're an ecosystem of different plants and animals. They all have to collaborate to grow. So that's our charter. We live in this world. We are building a very good ecosystem. And in the process, we make very good friends. Which is what I enjoy.
Chip Rodgers 05:01
That's great. I love that analogy, the coral reef is a really good analogy because every piece in that I had a saltwater tank many years ago and you always had to be careful about which fish you put with and all those things. So you're incubating a lot of those interactions and putting the right pieces together to create business together.
Manish Harsh 05:33
Summarized it right
How do Nvidia's partners use Nvidia's platform for innovative technology and end-use cases?
Chip Rodgers 05:38
Tell me a little bit, you mentioned ISVs, a lot of your partners or ISVs, what are the kinds of technology and end use cases and end customer use cases that those ISVs are typically working with, that they're provided to customers on using Nvidia as the platform.
Manish Harsh 06:05
We understand the large compute challenges. We go from workload. So let me take an example of recommender engines or large language models or cybersecurity threats. And I'm gonna detect different verticals, I can keep going. But we have identified the ones which are true candidates for leveraging in media technology, not only the hardware GPU, but the software as well, we are a full stack company.
So we have built AI workflows, because we understand them, we have built our resources and tools, which can be leveraged by developers and not waste time in reinventing the wheel, take it out of the box and start building their applications. Now, saying that it's not a full solution, they have to still work on top of it, but we make their life easy by not going at a very low level. So these workflows could be like, take an example of cybersecurity. It's a great use case, a strong candidate. But there's a lot of details into it, which we have already dissolved.
All what they have to do is if the SDK, understand, now, they may be building a full stack architecture with help of third party software sitting in middle which could be ML ops. One of these is important because different personnel will interact with each other and collaborate through this kind of software, which is today, not one single solution.
Companies stitch it, whether for data science and for training and development could be different and for deployment, inferencing could be different. We want to ensure that whatever tools they are stitching, they are able to leverage the power of the GPU and our software from the ISV because we are not interacting directly with the end developer many times.
Chip Rodgers 08:19
You've put a layer in between that is build for developers, at the ISVs that are providing that top level end user and user interface, and ultimately up to the customer.
Manish Harsh 08:37
Correct. So we do have open source, we have a large open source community, we probably have more than 100 Plus soft SDKs, which are all available for you. But then there is an enterprise version of it. So if you're in a mission critical production grade on development, you can go with Nvidia AI enterprise, which is our enterprise offering, in which we stand behind our software and support them in real time.
How does Nvidia work with partners to understand their requirements?
Chip Rodgers 09:10
How do you work with partners? How do you understand the requirements that partners not only end customers. So you have to think to layer two levels deep you got, what are the end customers looking for? But then what are your partners looking for in terms of making their life easier? How are you working with them on a regular basis, you hear challenges that they have, and feeding that back into the product teams or how does that affect you? How do you think about that from a partner standpoint?
Manish Harsh 09:50
That's a great question. This is where it becomes technical for us. We are a relatively large team. And we have assigned ourselves to a certain set of partners based on our expertise and skills, we all play a role in a matrix organization. We have weekly, bi weekly, or monthly cadence with the partner, key executives or technical architects to ensure that we understand where they're going, where they're stuck, where they need help, what kind of customer challenges they're dealing with. And those cadence check-ins with them help us understand.
We also do a lot of road shows.We will go do roundtables, execute roundtables with a partner and their customers to have more in person informal conversations about what they may be trying to do, and we can help them. So we do have these kinds of technical things, which each individual habits their own way, but mission is the boss, you want to understand the customer pain point, help the ISV. Understand Nvidia tech so well, that they can enable it the way they want to.
Chip Rodgers 11:14
Which then shows off your product makes it easier for them, but then also shows hopefully shows off the NVIDIA platform in a positive way as well.
Manish Harsh 11:30
It is positive so it always shows as positive.
Chip Rodgers 11:38
Well, speaking of road shows, Nvidia has an event coming up next month?
Manish Harsh 11:44
Thanks for bringing it up. It's a GPU tech conference, GTC it's our pride. It is amazing. It's just not the basic conference. It is a resource pool. It is knowledge for our developers. It is targeted to developers at very technical conferences. It's like, people just love to be there. I believe we have a really great attendance there. I don't know the numbers, but it's coming on March 20. It's virtual this time, because of the reasons. I would really encourage whoever's watching that not to miss the GPU tech conference GTC.
Chip Rodgers 12:30
I'm sure you have, in addition to outbound education, training, those kinds of things. You're probably planning some of those roundtables and feedback sessions and some of those things with customers and partners as well.
Manish Harsh 12:49
Yes, absolutely. We have. We call it a Deep Learning Institute, we have live training, workshops. Probably 800 Plus sessions in and during that one week of event. You'll have a lot of panels even though I'm actually hosting a lot of partners. We just love to see and it will all be available or once you're registered online.
Chip Rodgers 13:16
It'll be good. It'll be fun.
Manish Harsh 13:20
Watch parties happen around it. So that's an amazing moment for us.
Chip Rodgers 13:26
People getting together locally.
Manish Harsh 13:30
Yes, they get into their offices, and they don't watch parties. That's amazing. All the partners host their own watch parties.
Chip Rodgers 13:37
That's a great concept, then the partners are the collection point for their own people, and customers. And then they can connect together all around the video platform.
Manish Harsh 13:53
In the end, it just shows the passion and that's a true testimonial to our ecosystem. That's Nvidia's ecosystem. And it's a vice versa to be a part of their ecosystem. I'm seeing watch parties happening in Israel, Telavi, Germany, in Asia, India.
Chip Rodgers 14:18
That's Fantastic.
Manish Harsh 14:20
Jensen’s keynote is really good. It's a treat to watch because he will give his vision. It's his announcements on the products, and he will show the bot to all the partners and customers. So that's a must watch.
Nvidia's vertical solutions in AI/ML and cybersecurity
Chip Rodgers 14:36
So let's dive a little bit because I think it's really interesting. Some of the use cases are really interesting. Maybe we could talk a little bit about some of the vertical solutions and you mentioned some cybersecurity and some of the AI and ML solutions, where'd it be, tell me a little bit more about some of those vertical solutions where in Nvidia really shows off?
Manish Harsh 15:19
We are on a mission to start thinking vertical, as we mature in any use case. So whether it's automotive and healthcare, telecom, manufacturing, robotics, data center, energy, oil and gas. They are all unique. Although you can say that one solution fits all, but that's not you need to double click, into understanding the business before you say here is a solution for you.
So, for us, it's an edge in a data center. We want to do things end to end, from data collection point to data compute for the AI purpose. So we are building solution recipes, and frameworks, and SDKs for each vertical. That's how we are progressing. I would like to say Automotive is the most mature where we have an entire Nvidia drive platform. And we understand we have our own autonomous vehicle riding on the roads. Similarly, we just keep going and in each industry, there are dedicated teams. We have ecosystems around each of the verticals. And we have solutions, the most in demand use cases for them have been all identified based on the workload. And we have an offering for each of them. But that's a longer conversation. That's something I've just put a template on.
Chip Rodgers 17:05
So you're actually creating separate SDKs for industry use cases. Tell me about that. What drove Nvidia to do that was to say, okay, look, if we're talking automotive, then the SDK needs to look like this. If we're talking about telco, maybe it looks a little bit different. What pushed in that direction, versus having a single platform where you operate from?
Manish Harsh 17:38
GPU is GPU. But it has different form factors, it can be used for multiple scenarios, we wanted to make life a little easier for the developers, we all have our developers, we have a 3 million developer community and it's growing, we want to make sure that we are helping the developers not have to scratch their head on how to use it.
So I won't say the SDK solves the entire problem for them, they still need to build their app on top of it, it's their recipe they need to, but we make the life easy for them. That's the idea. And that's the way we want to support our partners and customers as well. As our CEO says, our job is to make complex things simple. AI is complex but we want to simplify it to a point where they focus on their core strength.
Chip Rodgers 18:44
That's really interesting. So essentially, if you're then an ISV, that's in the cybersecurity space, basically, you sort of just strip out all the things that might apply if you're building autonomous vehicles or, you know, something else in a telco or retail space or something like that. It keeps it very focused for those developers.
Manish Harsh 19:12
And we will train it with the right data, which is in that industry, and we'll bring it to a point where we can stand behind it. This is very much close to what you want. But you know, now here is a container, here's the resources, here is a model, pick it, train it with your own data and you know, start going to life. So this is all the development phase moment you get into enterprise or you get into a production level. We have support offerings for them.
Overview of Nvidia's partner programs and how they are tailored to specific partners?
Chip Rodgers 19:50
Talking again about partners, Nvidia has a lot of them. You've mentioned some areas like this, but some really strong different programs for partners. Tell me a little bit about those , you have a startup program and several others accelerate to think and a number of different programs. Tell me a little bit about what those are and how they're optimized or focused on specific partners.
Manish Harsh 20:37
Let's say, Hello World, there are not many programs, internally, we do just fragments so that we can serve them better. There's Inception program, which is for startups and disruptive technologies who can leverage your end, we are not picking winners, we are wanting to help everyone. So Inception is majorly for startups, and want our help to grow.
We have an NPN, which is Nvidia Partner Network, which is more for all the kinds of partners whether ISVs, global system integrators, RSIs or solution companies or could be in that paradigm, we do have a special program called AI accelerated program where companies who are way more integrated into our technology and we validate them.
They enable Nvidia tech from inside their UI or whatever their software is. They are validated and certified against our common enterprise offering. Those are the three key programs, but there are plenty more programs run by vertical teams, but they all roll up or tie into each other at the end of the day, between these three to five programs. But it's just a way of internal mechanism. But for the outside world, we have one developer program, we have one partner program, one Inception program.
Chip Rodgers 22:12
It's interesting the organization and video is just always very focused on partners and makes sense because you're not going to the store and picking up an Nvidia CPU off the shelf. There's always something that's a part of it.
Manish Harsh 22:40
We have been keeping our statements very true and very socially driven in many things, the gaming community is very big in number. Developers are the community. So we are evolving as we are able to grow. The good part is we are all very well stitched internally. In company everyone understands the mission. We have a big fleet of developer relations, people who are kind of evangelists and talk, understand, build this kind of network with the partners, which we call second party developers or third party developers, however you would like to say. It's a fun journey. I like that we understand our road very clearly.
How does Nvidia manage joint roadmapping and partner collaboration with product teams?
Chip Rodgers 23:37
We've touched on this but maybe we could dive a little bit deeper on the process that you go through with partners to build solutions together? They're there, they've got something that might be interesting around a specific vertical or some new use case or capability that they're thinking about?
Do you talk about joint roadmaps and how tightly are you connecting the partners to the product teams, so that you're staying in your own lanes? And maybe if you could talk Manish a little bit about how that works, and how you orchestrate that activity?
Manish Harsh 24:37
Depending on what the problem statement is, we will go if it is a really great statement and very useful to be solved. And it's really a game changer or it's just important. Let's put it that way. We will go lengths to collaborate. Now depending on the type of partner and bandwidth, if let's say a larger partner would want to build a center of excellence to showcase we will give them the reference architectures and entire infrastructure required for that we will support because we build supercomputers with a just a software level of a lot of software companies, if they need some help in understanding the calculating the compute infrastructure or the networking part or the storage required to solve a setting, we will do that because we have all that knowledge sitting in our company.
We want to ensure that they understand it. So, there are so many parallel motions going on. Again, it's very individually driven. We have account executives and managers, developer relations people, solution architects. We all go as a team and we'll help automation.
Chip Rodgers 26:03
We actually had a question come in from R Ravindra. Comments and questions. So agreed every industry ML ops use cases are vertical specific. What we've been talking about in the semiconductor/chip dev space SDK use case offering from Nvidia through his partner network.
Manish Harsh 26:32
The question is not very clear but my understanding is he's asking about if you have a specific SDK offering from the Nvidia partner network. Yes, we do. We actually if you go to the Nvidia ML ops partner portal page, you will find a list of partners who are validated, certified, and we stand behind them, because they have elapses in class. And we are building what we have as we were talking.
We are building this ecosystem last year, it's a relatively new area, we are looking at around 300 plus companies in that landscape. So the way we are going to do it, we will enable and support these ML ops companies to ensure that they have the right SDK that is integrated to their software. And then we'll promote them.
Advice for working with Partners
Chip Rodgers 27:27
Excellent. Manish, this has been fantastic. Maybe we could round out our conversation, I love to ask partner people because we have partner folks that are in our audience. What are some things that you picked up along the way? When you're working with partners, and you have a real specialty and working with developers, some words of advice that you picked up along the way that you would want to share?
Manish Harsh 28:06
When I'm wearing this t-shirt, black and white, there's some things with the partner relationship that has to be very black and white, you should understand each other's motivation, you should understand each other's bandwidth. And not every partner has the same size and shape to dance along with you for long. So understanding their strengths and limitations, it will help connect and support each other.
We constantly keep thinking how to add value to our partner, if you're not adding value, you're not doing justice to the friendship. So, we treat partners very much like friends, and have to be executive level alignment, technical understanding of each other's tech skills and the roadmap, some partners may not have any alignment with you, that's okay. You don't have to force yourself on each other. But you'll find your players and then you go together, there's no winners, we all win together. And we will fail together if we don't.
Chip Rodgers 29:24
I think that's a good point to end on and I would also bring it back to the very first thing that you said, which is that your job is to make friends.
Manish Harsh 29:34
Truly and it has to be from the heart you just cannot be fluffy in this. Otherwise you will not scale. The ecosystem will scale only if you are true to each other. And you constantly keep adding value to the relationship and the partnership. You have to think about how we can make it better. Could be GTM. It could be business to have, it could be blogs. webinars, executive roundtables, helping end customers doing POCs, there's so many tactics, you pick your best and whatever is your passion add value. That's my recommendation.
Chip Rodgers 30:20
Good words of advice. Thank you. Manish, this has been fantastic. I really appreciate you sharing your thoughts and it's a hot topic these days just around with all AI and ML and chat GPT is putting it on everybody's radar. It's become a culture. This is pretty powerful. Maybe a little scary.
Manish Harsh 30:51
And you've been interviewing people in this domain. I would love to learn someday because you get information from multiple people. And so learning is always open to new ideas.
Chip Rodgers 31:05
Well, this has been terrific. Manish, thank you again. And I really appreciate you sharing your thoughts and your advice and, and ideas.
Manish Harsh 31:19
Thank you, Chip, for having me here. I hope it was helpful.
Chip Rodgers 31:22
Very much. Thank you all for joining us for another episode of the Ecosystem Aces and from Manish Harsh I'm Chip Rodgers and signing off and we'll see you again next Friday for another ecosystem Aces thanks everybody.
Links & Resources
- Learn more about how WorkSpan helps customers accelerate their ecosystem flywheel through Co-selling, Co-innovating, Co-investing, and Co-marketing.
- Subscribe to the Ecosystem Aces Podcast on Apple Podcast, Spotify, Stitcher, and Google Podcast.
- Find insightful articles on leading and getting the most out of your partner ecosystem on the WorkSpan blog.
- Download The Ultimate Guide to Drive Co-sell Partnerships!
- Download the Best Practices Guide for Ecosystem Business Management
- Download the Ultimate Guide for Partner Incentives and Market Development Funds
- To contact the host, Chip Rodgers, with topic ideas, suggest a guest, or join the conversation about modern partnering, he can be reached on Twitter, LinkedIn, or send Chip an email at: chip@workspan.com
