AI Corporations and Communities in Africa with Karim Beguir & Muthoni Wanyoike
On the podcast today, we have two more fascinating interviews from Melanie’s time at Deep Learning Indaba! Mark helps host this episode as we speak with Karim Beguir and Muthoni Wanyoike about their company, Instadeep, the wonderful Indaba conference, and the growing AI community in Africa.
Instadeep helps large enterprises understand how AI can benefit them. Karim stresses that it is possible to build advanced AI and machine learning programs in Africa because of the growing community of passionate developers and mentors for the new generation. Muthoni tells us about Nairobi Women in Machine Learning and Data Science, a community she is heavily involved with in Nairobi. The group runs workshops and classes for AI developers and encourages volunteers to participate by sharing their knowledge and skills.
Karim Beguir helps companies get a grip on the latest AI advancements and how to implement them. A graduate of France’s Ecole Polytechnique and former Program Fellow at NYU’s Courant Institute, Karim has a passion for teaching and using applied mathematics. This led him to co-found InstaDeep, an AI startup that was nominated at the MWC17 for the Top 20 global startup list made by PCMAG. Karim uses TensorFlow to develop Deep Learning and Reinforcement Learning products. Karim is also the founder of the TensorFlow Tunis Meetup. He regularly organises educational events and workshops to share his experience with the community. Karim is on a mission to democratize AI and make it accessible to a wide audience.
Muthoni Wanyoike is the team lead at Instadeep in Kenya. She is Passionate about bridging the skills gap in AI in Africa and does this by co-organizing the Nairobi Women in Machine Learning community. The community enables learning, mentorship, networking, and job opportunities for people interested in working in AI. She is experienced in research, data analytics, community and project management, and community growth hacking.
Cool things of the week
- Is there life on other planets? Google Cloud is working with NASA’s Frontier Development Lab to find out blog
- In this Codelab, you will learn about StarCraft II Learning Environment project and to train your first Deep Reinforcement Learning agent. You will also get familiar some of the concepts and frameworks to get to train a machine learning agent. site
- A new course to teach people about fairness in ML blog
- Serverless from the ground up: Building a simple microservice with Cloud Functions (Part 1) blog
- Superposition Podcast from Deep Learning Indaba with Omoju Miller and Nando de Freitas tweet and video
- Instadeep site
- Nairobi Women in Machine Learning and Data Science site
- Neural Information Processing Systems site
- Google Launchpad Accelerator site
- TensorFlow site
- Google Assistant site
- Cloud AutoML site
- Hackathon Lagos site
- Deep Learning Book book
- Ranked Reward: Enabling Self-Play Reinforcement Learning for Combinatorial Optimization research paper
- Lessons learned on building a tech community blog
- Kenya Open Data Initiative site
- R for Data Science GitHub site and book
- TWIML Presents Deep Learning Indaba site
Question of the week
If I want to create a GKE cluster with a specific major kubernetes version (or even just the latest) using the command line tools, how do I do that?
Where can you find us next?
Our guests will be at Indaba 2019 in Kenya.
Mark will be at KubeCon in December.
Melanie will be at SOCML in November.
Transcriptshow full transcript
[MUSIC PLAYING] MARK: Hi, and welcome to episode number 152 of the weekly Google Cloud Platform Podcast. My name is Mark Mandel. And as always, I'm here with my colleague, Melanie Warrick. How you doing, Melanie?
MELANIE: We're in the same room.
MARK: We are in the same room.
MELANIE: I don't know when was the last time we were actually in the same room.
MARK: Probably about three or four weeks ago.
MELANIE: It was a long time ago.
MARK: It was. Pretty excited this week, you're bringing yet another one of your amazing episodes or interviews that you had while you were in Africa.
MELANIE: I am. This is our last interview, actually, that we captured while I was over in Cape Town, specifically Stellenbosch, at the Deep Learning Indaba. And we specifically met with Karim and Muthoni who both are out of InstaDeep. And they talked to us a little bit about their corporation as well as just, in general, AI corporations out of Africa and communities that are being built. Because Muthoni, in particular, has a community that she's established out of Nairobi.
MELANIE: So we get into that, which will be great. But as always, we start out with our cool things of the week, and we're going to end with our question of the week. And the question, which is coming all the way from Mark, which is, if I want to create a GKE cluster with a specific major Kubernetes version, or even just the latest, using the command line tools, how do I do that?
MARK: How do you do that?
[DRAMATIC ORGAN MUSIC]
MELANIE: I don't know. But hopefully you know--
MARK: I hope so.
MELANIE: --so that you can tell me how we do that. All right, so cool things of the week. We're going to kick it off with an article we both found. And this was-- actually Mark found. And this is this article that's titled "Is There Life on Other Planets?" Google Cloud is working with NASA's Frontier Development Lab to find out.
So it's really interesting actually. It talks a little bit about these two approaches that they've broken down into researching the biologies of distant planets. And one approach is, in particular, they're modeling the possible atmospheres that could exist on distant planets.
And they analyzed more than, like, 270,000 simulated atmospheres on Compute Engine. And this modeling that they did, they actually have code that they've shared on GitHub. It's called PyAtmos, and there's a link in the blog that you can check out.
And then the second approach is that they're then generating an actual spectral data set of over three million rocky terrestrial exoplanets with machine learning.
MARK: Oh, my god.
MELANIE: So apparently we don't have enough data yet. But we have a ton of visual images that have been captured by different satellites that exist. And then they've done a significant amount of simulating as well to try to help come up with good models to identify rocky terrains that would potentially have life.
So that data set, they're planning to release sometime in the near future. And you can keep an eye out for it. But you should check this blog out. It's interesting and just hearing about what's going on in that space and how machine learning technology is playing a role.
MARK: Cool. And my next cool thing of the week combines actually both of us, games and AI, which I think is really cool.
MELANIE: It's special.
MARK: It is special. One of the community members-- and I'm going to apologize. I'm going to mess up your name. I'm going to assume its Gema Parreño. It's probably close.
MELANIE: Good job.
MARK: I'm going to go with that. They are a data scientist. They've been doing amazing stuff with a StarCraft II learning environment project. So if you're not familiar with that, that's a project that comes out of DeepMind. PySC2 is DeepMind's Python component of the StarCraft II learning environment basically. It exposes Blizzard Entertainment's StarCraft II's machine learning API as a Python environment. Basically, you can build intelligent agents that play StarCraft II, which is kind of awesome.
MELANIE: It's all part of this effort that's going on right now, like DeepMind had worked on AlphaGo. And that's part of their space, Dota 2 out of OpenAI, exploring reinforcement learning algorithms in the machine learning space to be able to play these games and have the machines play these better than humans, in essence. But this is a place you can experiment with and tools you can experiment with.
MARK: So yeah, the cool thing of the week is not just that. Because that is cool. But the thing that Gema's actually done is if you're listening to this, and you're like, that sounds great. I want to play with it. They wrote a CodeLab that you can go through to learn about the environment, how to set it up, how to get it going, and basically start building and running your own agents, which is super cool. It's really cool as well, in that-- well, you do need StarCraft, the game, but you don't need a license to actually set it up and run it, which is really nice.
MELANIE: Yeah, I like how I'm able to get you over to the machine learning side with the games. Another cool thing of the week we want to mention is there's an additional module that has been added to the Machine Learning Crash Course that's out there. And this module, in particular, is focused on fairness in machine learning. So you should check this out if you have any interest in this.
I brought this up, I know, a lot this year. And we did a podcast a while back about machine learning bias and fairness. And so this is a nice module that helps you explore that in looking at different types of human biases that can manifest in training data. And it helps provide strategies to identify them and evaluate their effect. So you can see that in the Machine Learning Crash Course and check it out.
MARK: Nice, awesome. We have one of our wonderful developer advocate teammates, Martin Omander. They have written a "Serverless From the Ground up-- Building a Simple Microservice with Cloud Functions." This is Part 1, apparently.
They take a sort of real-world approach of coming up with a company that needs to do a certain thing, in this case, listing and viewing docs, and then taking that and saying, OK, how can we do that with Cloud Functions? What would that flow look like? How would that work with, say, in between talking to different teammates and how they would work together on this? It's actually a really nice step by step instructional on how to build up these sort of microservices.
MELANIE: And also, last thing of the cool thing of the week is that while I was at Deep Learning Indaba, I met Emily [? Mueller ?] and [? Ribone ?] Miraba who are building out their own podcast called Super Position. And they did this recording with Omoju Miller and Nando de Freitas who talked about machine learning, talked about the conference. It's a great podcast and a great video that they included. So we're going to include a link to give them a shoutout in our podcast for this week.
MELANIE: Yeah, it seems appropriate, considering we are also talking about Deep Learning this week. All right, Mark, I think it's time for us to get into our interview.
MARK: Let's go do it.
MELANIE: On today's podcast, I'm excited that Karim Beguir and Muthoni Wanyoike are both joining me to talk about InstaDeep and building communities in Africa. So thank you both for joining.
MUTHONI: Thank you for having us.
KARIM: Thank you.
MELANIE: We are actually recording this-- this is another one of our podcasts that we are currently recording out of the Deep Learning Indaba, which is currently happening in Stellenbosch, South Africa. So let me go ahead and get us started like we usually do, which is, Karim, can you tell us a little bit about yourself?
KARIM: Sure, so I'm half Tunisian, half French. I actually grew up in a small city close to the Sahara called Tataouine--
[PLAYFUL ORGAN MUSIC]
--like the sand planet in "Star Wars."
MELANIE: I was about to say, "Star Wars."
KARIM: Exactly. And literally, I grew up through learning, progressively made my way to graduate to Ecole Polytechnique in France and, ultimately, the US, where I was at the Courant Institute at New York University. And through the process, I always had this idea to give back, come back to Africa at some point to do something interesting and useful. And this has guided my choices to what is, today, InstaDeep.
MELANIE: Muthoni, tell us a little bit about you.
MUTHONI: Yeah, thank you. I grew up in a small town in Kenya called [INAUDIBLE]. And after that, went to school in [INAUDIBLE] and then started working in the city, Nairobi. I work at InstaDeep. I lead the team in Nairobi as well as our strategy of expanding into Africa and building talent across Africa.
MELANIE: OK, so as I mentioned, this is being currently recorded at the Deep Learning Indaba. I know we're not going to be releasing it exactly at the same time. But what has been really valuable experiences for you from this conference so far?
KARIM: I think the level of enthusiasm is fantastic. The quality of the speakers is really top-notch. I like to see this as a NIPS, you know, same quality and speakers that you would have at NIPS, except that NIPS has 5,000 or 6,000 attendees. And here, it's a much smaller audience. So I think it's an exceptional opportunity to really build AI communities in Africa. And we should be very thankful to the Deep Learning Indaba team for what they are doing and the changes they are making on the ground.
MELANIE: And Muthoni, what are some of the valuable experiences for you?
MUTHONI: For me, I think, having been part of planning the Indaba this year, the main takeaway for me has just been to see all the people that we've been communicating with for many months come together and to also see the quality of research that is coming out of Africa. And very interesting posters are being presented every day, as well as very interesting sessions, for example, to build on machine learning skills or how to write good paper skills and that kind of experience has been really good.
MELANIE: Muthoni, I know you are involved with actually putting together this conference. How did you get involved?
MUTHONI: It's a very interesting story. So my colleague, my co-organizer of the community in Nairobi, came to the Indaba last year. And the planning committee had already decided that they wanted to host the Indaba in Kenya in 2019.
When she came to me and shared the vision of the Indaba, which is to strengthen the participation of Africans in machine learning, I was sold immediately. Because it's what we do with the community in Nairobi. So then we had a chat with [? Shakir ?] and the rest of the team. We seemed to like each other, and we started the process of planning the Indaba this year.
MELANIE: And the rest is history.
MELANIE: Now you're in, and you're never going to leave.
MUTHONI: Yes, and we'll be hosting 1,000 people next year, so that should be exciting.
MELANIE: That is really exciting. And Karim, you're here this year because you're speaking. Can you tell us a little bit about what you're talking about?
KARIM: Yeah, so my session is going to be on Thursday about the life of a machine learning startup. And the goal is really to give some insights about what it takes to build a startup but specifically a machine learning startup and specifically in Africa. So looking forward to share some knowledge and also raise the awareness in the AI community about the opportunities that there are for AI in Africa.
MELANIE: What are some of the opportunities?
KARIM: I think we have tremendous talent. And one of the insights that I want to convey is that in today's world, it's actually possible to build a high-quality machine learning startup in Africa. All the sources of knowledge are there. The sources of mentoring, high-quality mentoring, are there now, with events such as Deep Learning Indaba. So really make the new generation aware of what's possible and share some of the enthusiasm and some of the experience that we've been through at InstaDeep.
MELANIE: And tell us a little bit about what InstaDeep does, how it started.
KARIM: So InstaDeep is an AI startup. It's an African AI startup which bridges the gap between advanced AI research and large companies that have real needs, real business needs. So the way we've done that is by developing our own research but being able to explain and convey to the communities, to the companies how advanced AI can benefit to them. So that has been our experience.
And what we've discovered by doing that-- also, the project started in North Africa in Tunisia, started as a bootstrap, literally. My co-founder, Zohra, and I started with two laptops and a lot of enthusiasm. And what we've discovered is that it's actually possible to bring promising African talent to competitive levels in AI, to be able to add real values to companies in Africa and beyond.
That's the main takeaway, and that's also my main message. We have to be optimistic about AI in Africa. The opportunity is tremendous. And I think this is our agenda, all here to convey that message and make sure it's widely spread.
MELANIE: It's definitely not a hidden agenda, for sure. Well, in terms of InstaDeep, when we were talking offline, you were telling me that it's in five different locations.
MELANIE: And where are those locations?
KARIM: So we started initially in Tunis in Africa, and now we have three locations. We're operational in Kenya. That's the team that Muthoni leads. We are also operational in Lagos, Nigeria. And we also have offices in Paris and London, so five locations and looking forward to continue to expand our presence in Africa.
MELANIE: And Muthoni, how did you get involved?
MUTHONI: Through my community work. Karim had visited Nairobi because he is a Google Launchpad Accelerator mentor. So Nairobi is quite the city, and you just fall in love immediately. So he was interested in setting up in Nairobi and was meeting with people within the community. So that's how I met with Karim and was sold on his vision of growing AI in Africa and providing opportunities for a billion people to work on exciting projects. And yeah, the rest is history again.
MELANIE: The rest is history. Now you're working with a group. In terms of-- you had mentioned about the Google Launchpad Accelerator. I know you're also a GDE, a Google Developer Expert.
MELANIE: I also know-- we talked about this-- that you're using Google Cloud products as well. What are some of the products that you're currently using at InstaDeep?
KARIM: TensorFlow is our main platform in AI. It's a tremendously powerful tool. I think people do not realize how powerful this tool is, and we're really grateful to have that free access that Google has provided. So in my experience, as also an ML GDE, I've come to appreciate all the work that Google is doing, particularly, for example, products such as the Assistant are tremendously powerful and make advanced AI available to all startups.
I'll mention another one which I think is extremely impressive, which is Auto Machine Learning, AutoML. That is a groundbreaking product. It effectively enables startups all over the world, including in Africa, to have access to high-quality AI without needing extended teams.
So participating to launch product accelerator Africa with Google by helping build an ecosystem has been also a tremendous opportunity for me to meet great people. And it's through those channels that I have come to meet with Muthoni and also been convinced that now is the time to make a difference on the ground in Africa. So it has encouraged me and acted as a catalyst to the vision I initially had.
MELANIE: And you were mentioning about communities, Muthoni. I do you want to talk a little bit about-- there's a community, in particular, that you run, which is the Nairobi Women in Machine Learning. How did you get involved with that community?
MUTHONI: The community with [INAUDIBLE] two years ago. I had just graduated from my undergrad and was working in government, the Kenya Open Data Initiative. And there was a lot of buzzwords around big data, AI, and all that. And there was a lot of expectations in our scope of work.
So we were learning a lot of new skills. But I felt the need to connect with people who are working in a similar space so we can share ideas or resources or just know what is happening around the world together. So that's how I learned about the Women in Machine Learning & Data Science community and reached out to them. And they were happy to help us get set up in Nairobi.
And then after that, we did a call for volunteers who are interested in helping build the community. So two years later, we run a lot of workshops, from introductory classes on how to use Python and R to more advanced classes on TensorFlow and other advanced AI concepts. And it's very interesting, the range of people attending the lessons. For example, we have university lecturers coming to learn Python or a 13-year-old girl coming to learn how to program. So it's been a very interesting experience building that community in Nairobi.
MELANIE: How large is the community right now?
MUTHONI: We have close to 1800 people on Meetup. We probably have met 700 of the people. So it's a huge group. And how we run things is that it's very volunteer-lead. So a lot of people will come to us and then see a need in the community. And we let them champion that as a project within the community.
MELANIE: That's great.
MUTHONI: One interesting one that I would mention is we did a master class celebrating the International Women's in March. We did a master class introduction in Python. And the ladies attending were very, very interested in it. And two of them volunteered to grow that class. And now they are doing a course across four months and have a class of about 30 ladies in Nairobi and in Kenya doing the course by themselves remotely and then giving feedback every week and results and sharing knowledge and skills.
MELANIE: What do they use to help work remotely together?
MUTHONI: They use Slack. So they coordinate class in Slack. We took the "R for Data Science" workbook by Hadley Wickham and then divided that across and made that our curriculum spread out over four months. And then we used Slack to coordinate work as well as GitHub.
Because a lot of the people, also, were not using GitHub. And that really counts when you're, for example, applying for a job. So now they can upload their projects there, and they're able to showcase their work when dealing with recruiters, for example.
MELANIE: I know. I'm pretty impressed at how important GitHub has become in the space. And when I talk to people and I tell them all the time-- and even here, I was talking to people about how important it is to publish on GitHub, to not wait for it to be perfect, to just get it out there. Show that you're working on something. So do you collaborate with other women in machine learning communities or other communities in Nairobi or outside of Nairobi?
MUTHONI: We do. So for example, just out of us having the community in Nairobi, now we have a community in Lagos of Women in Machine Learning. We also have worked with the few women taking the data science class in Uganda. And just at the Indaba, I've met very many ladies who are very excited to have similar communities in their countries, so very, very exciting times for building the skills of women in Africa for machine learning.
MELANIE: It sounded like you said it was on Meetup. So if people want to explore or check out the group, that's where they can find it?
MELANIE: Well, Karim, what is community? I mean, we talked a little bit already about this. But what does that mean, to you, in terms of these types of communities and your company and the interest that you have in AI?
KARIM: So I think that's very important, to help communities understand the AI opportunities. So a lot of what we do is actually trying to bring that know-how and helping people focus on what really matters. I would say, today, the problem is not the lack of information. There's maybe an information overflow.
So a big part of what I do, as a mentor-- and we also organize TensorFlow Meetups, very active. We organize a hackathon. A month ago, I was in Lagos organizing a hackathon. I was really impressed by the talent and the passion of the young students and engineer there.
So a big part of what we do is raise awareness, break a sort of mental idea that, oh, you cannot be world-class in Africa. Actually, as an African AI startup, you can compete with the best. And we had really, very constructive exchanges here at Deep Learning Indaba with Nando Freitas and David Silver from DeepMind. And their feedback was really, very positive, and this encouraged us to continue to share the message all through Africa that AI is a real opportunity, that you can be world-class in AI. All what it takes is positivity, willingness to learn, energy. And I believe, in the continent, we have a lot of that.
MELANIE: And any advice or recommendations for those companies that are considering looking at having their own offices in parts of Africa?
KARIM: I think the key thing in AI and in many other things is focusing on the people. So for example, at InstaDeep, we do not have a specific plan that we should be, for example, having an office in a particular country or city. It is really driven by the quality of the people.
When we feel we have the right partners-- and we're so excited to have Muthoni in Kenya. We have a great machine learning team also in Nigeria. So when you have the right team, the right people, you can build. And that's my main message, and this is, I think, particularly relevant in Africa. It's about investing on the right people who share your vision, who share your enthusiasm.
So my message to companies looking potentially at hiring African talent is try to look actively for the top talent and, most importantly, for the people who have a willingness to learn, a willingness to challenge themselves to get to a certain level of competitiveness and collaboration within the team. So I've been really surprised by how quickly we've met some of the right people. And at the end of the day, in AI or anything else, it's people-driven.
MELANIE: Very much so. I agree and appreciate that too, from the standpoint of talking about the community-building, talking about the corporation having that opportunity to explore these communities and see where the talent is and then also giving back. And giving back in terms of the training is phenomenal.
I know we talked about this earlier already. You were touching on some of the tools that are being used. You know, when you bring in your talent, it sounds like you're doing some of your own training as well and helping people grow. Is that correct?
KARIM: Yes, absolutely.
MELANIE: Yeah, and so, what are some of the key skills, in particular, that you're trying to build out?
KARIM: So we believe that it's possible to take a motivated young student or engineer, let's say undergrad or completing their grad studies, and get them very quickly working on AI research problems, which would be considered PhD-plus. I think that's been one of the fascinating surprises I've had building the company is to see that this is actually possible. We had great success stories, some fantastic women students we had on the team. InstaDeep really believes in diversity and making sure everybody has an equal opportunity.
And we've been really surprised by success stories. We've had interns push the state of the art of work that's been done in Silicon Valley. So this is, I believe, an exciting time. And the key message I want to convey is do not believe the AI opportunity is something removed from you. If you're truly motivated, if you truly want to learn, there is a way. And there are mentors available to help, whether me or all the amazing people at Deep Learning Indaba, to make that opportunity a reality for you.
MELANIE: What are some of the resources that you find more valuable for teaching and for helping others train? Because I know we talked about Hadley Wickham and how his book has helped with some of the training that you're working on with the Nairobi Women in Machine Learning. Karim, do you have any ones that particularly come to mind that you find very valuable?
KARIM: Yeah, absolutely. There are a couple of books which we found to be tremendous resources in practice. The good news is that they're for free. All that it takes is go and click a link, and you'll get them. Download the PDF, print it, and, here, you're in business.
So I say, one major book, if you're interested in deep learning, is the "Deep Learning" book by Ian Goodfellow, Courville, and Bengio. That is a fantastic one, and I love to say it's a fantastic one because it's a real test. It starts really easy, and the difficulty almost increases exponentially chapter by chapter. So if you're motivated, take the Goodfellow, and make sure you finish it. You'll be a long way into a positive and successive career in AI.
MUTHONI: I'd like to mention something interesting that motivates me at InstaDeep in terms of growing talent. So we also have very experienced researchers working at InstaDeep as well. And they somehow are paired with our junior engineers, and they are able to work together very well. So that really helps our engineers grow at a really, really fast rate, in terms of their skill.
MELANIE: That's a common challenge, too, with most companies is having the ability to bring in junior talent and having them be able to grow. And the fact that you're actually mentoring and setting up that kind of relationship-- some companies are doing that. Not everybody, unfortunately, is able to execute on it. They're not doing it. They're not executing on it.
KARIM: And I think Muthoni's touching an important point. At InstaDeep, we make sure that all projects are cross-regional, so different teams have to collaborate. We've experienced this, in particular, with the teams we have also in Europe. We find it to be a key component.
Everybody gets the same access to resources. Everybody gets to share and collaborate. And we've had tremendously interesting experiences with this. African talent can pick it up really quick. And we've had cases where, really, the innovation is driven from Africa. And we're very proud of that.
MELANIE: How so?
KARIM: So we've had, for example, research projects. I'll mention a recent article we've published called "Ranked Rewards," which is effectively taking the work that had been pioneered by DeepMind, and David Silver in particular, around two-player games. We found-- this is very interesting. It's a massive breakthrough. Actually, technically, AI was invented by Alan Turing to solve chess. So you could say [INAUDIBLE] is really like a moment in time. And we were very excited by those breakthroughs.
And we tried to think about, how could this breakthrough apply to real use cases, real businesses, real situations, whether it's around transportation or logistics-- how to make those useful. This has led us to breakthroughs, which were pioneered, in particular, by our North African team, which got us great reviews, great feedback, and ultimately building next generation products that will benefit all team, all communities and beyond.
MELANIE: That's great. So off of that point about working remotely-- and you had this great example in terms of what was being driven from Africa-- what are some of the biggest challenges that you've had working with remote teams working together?
KARIM: I think it's important to make sure people communicate and collaborate. So some of the challenges are more like practical. For example, when we organize an off-site, make sure you're going to get the visas. Because unfortunately, it's very hard to travel around Africa.
So problems are practical in nature. But if you are really motivated, you could go beyond them and create a culture of collaboration. So at InstaDeep, we really believe in that. We make sure at least twice a year the whole global team meets in a location to exchange and share directly and build up an inclusive culture.
So software is, I believe, the way to go in Africa, due to some of the difficulties around moving people, moving goods. So we still struggle sometimes with visas. And Muthoni about this. We had to fight to get her in [INAUDIBLE].
But in startup, like in everything else, it's about motivation. If you really have the motivation to go beyond difficulties, ultimately, in today's world, you can. And the point I want to also make is that you're not alone. If you're an African machine learning startup, you have the chance to access to tech giants who are there to help you to develop the ecosystem. And I found that to be really transformational.
This also has guided me towards the work I do with Google as a machine learning Dev Expert. There are opportunities out there. So don't believe that you're on your own; you're not. There are lots of people who are super excited about the potential of the continent and will actually lend a hand and help you. That's been our experience, and if we can do the same and share that feedback and that positive feedback, I think we've gone a long way.
MELANIE: I see you nodding, Muthoni. Anything else you wanted to add?
MUTHONI: No, I just agree that it's quite difficult to move around the continent. But it's interesting to see the efforts towards just making things happen. Because, for me, representation is very important. And sometimes just this limit on moving-- for example, we had colleagues who were to attend the Indaba and couldn't get their visas.
That means it blocks a part of them accessing resources and people who would have been key to their next phase of life. But with technology, we are able to bridge some of these problems. And it becomes an easier way of life.
MELANIE: I mean, I love the fact that you're talking about particularly how all of your teams have the same resource access and that that's what you fight to achieve. But I also very much value the fact that you're trying to make sure that all these teams work together. I've worked with remote teams as well. And I feel like I'm stronger when I work with people who are from different backgrounds, different mindsets. And it sounds like that's a large part of what's driving this and also what you experience as well, correct?
KARIM: Yeah, you have to create a culture of inclusiveness and empowerment. All our teams are treated the same no matter where they are. We value ideas. We value intellectual curiosity. So we've had cases where somebody was supposed to work on something that turned out to be brilliant at something else. We love that kind of positive surprises. And if you create an environment which is positive, which pushes people up and rewards constant learning, intellectual curiosity, you are bound to have good surprises.
MUTHONI: People's motivation to do things cannot be overstressed. When people have their own personal motivation to work on projects, then it just makes, for example, even the small challenges of maybe time because you're are working across different time zones-- but when you are motivated to work towards exciting challenges, it makes some of those things-- you easily overlook them, I think.
MELANIE: Well, anything else that we didn't already cover that you wanted to touch on, that you wanted to emphasize, whether it's about InstaDeep, about building communities, about Deep Learning Indaba?
KARIM: To conclude, I think it's really an exciting time for Africa. Because there is a transformative technology, AI, which is happening and will continue to happen for the foreseeable future. And at the same time, the resources are progressively there to make taking opportunities, building businesses, employing people a reality.
So we're super excited about this at InstaDeep and within the deep learning endeavor. It's a community of like-minded people who really sense the opportunity. So I think this is a historic opportunity for Africa, and we're super excited about it. And ultimately, it's about building all this together.
This is the message of Deep Learning Indaba. We build together. And that will make a difference. So my message to anybody who is listening is do not believe that this is beyond you. Be part of it. Bottoms up approach, ultimately, is the way to go. This is what will change Africa, and all what it takes is a bit of motivation, a bit of curiosity, and not being afraid to get started. Great things will happen.
MELANIE: That's wonderful. And that actually reminds me of something I heard you say the other night. Muthoni was saying yes, saying yes to things that are also things that feel like they're impossible or out there beyond you. Did you want to add anything else?
MUTHONI: For me, my experience has been that I like to do personal assessment. And sometime earlier this year, I realized that I prefer-- my comfort zone is lurking in the shadows and just doing stuff quietly and silently or on Twitter. So when a friend recommended the [INAUDIBLE], then, for some reason, it started something in my mind. And that is how I ended up, for example, accepting to join InstaDeep because it was a challenge that I wouldn't have accepted a few months back or being part of the Indaba.
So yeah, I think I would encourage people to say yes to opportunities that really, really scare them. Because sometimes comfort doesn't grow you as much so, yeah. And one last thing, we will be-- I'm very excited to host the Indaba next year. It's another huge, huge challenge but I'm sure it's going to be a fantastic year. So I look forward to having the Indaba in Kenya in 2019.
KARIM: And on my side, I'm very excited about the next Indaba, which I'm sure Muthoni is going to be a big part of organizing. And also, maybe something else I wanted to add is about machine learning in Africa. I think now is the time to prove to the world, as Africans, that we can raise-- that even companies out of Africa can be competitive, can stand on their feet and basically provide the value added that we need to make sure all communities and all economies are successful in the 21st century. That's what we're trying to do at InstaDeep.
And if I had a wish, it's to see, in the future, successful AI companies emerging from Africa with the help and collaboration of tech partners and all people of goodwill. But this is what we haven't seen yet and, in particular, InstaDeep, that's what motivates me every day. Can we prove, as Africans, that we can raise to the opportunity and build something great for all the communities and all the partners we work with?
MELANIE: Karim, Muthoni, thank you very much for coming on the podcast.
MUTHONI: Thank you for having us.
KARIM: Thanks a lot.
MUTHONI: Thank you.
MELANIE: Thank you again, Karim and Muthoni, for coming on the podcast and talking to us about AI, corporations, communities, InstaDeep, all the things. We really appreciate it.
MARK: Yeah, thanks so much. That was really, really interesting.
MELANIE: Mark, question of the week?
MARK: Question of the week.
MELANIE: If I want to create a GKE cluster with a specific major Kubernetes version, or even just the latest, using the command line tools, how do I do that?
MARK: This is something that I've run into, and I was--
MELANIE: Really? Did you run into this?
MARK: I ran-- no, I did. I really ran into this.
Right, so GKE runs very specific versions of Kubernetes. So, for example, 1.10 is available right now. So if I'm running on the GKE, I might be running 1.10.6. And then usually, there's a patch release, so, like, -gke.6 or .7 or .4.
And those can roll through pretty quickly. You may be, on one day two or three weeks ago, you're running 1.10.6-gke.1. And then if you come back a few weeks later or months later, it might be gke.4, and the old version might not be available anymore.
And you might have run scripts either through the command line-- or possibly even Deployment Manager, as well-- that are very specific. They might be like, OK, yes, I need this version, [? 1.10.7.gke-3. ?] And that's not available anymore, and now suddenly, all your scripts break. And it sucks.
[DRAMATIC ORGAN MUSIC]
And I've run into that as well, where I have to update the script every time I want to create a cluster, either with gcloud or Deployment Manager. But you don't have to, which is super nice.
MARK: It's amazing.
MELANIE: It's amazing.
[CELEBRATORY ORGAN MUSIC]
MARK: So the really nice thing is you can actually specify a general cluster version. So say, for example, I want to create a version just like a 1.10. I'm like, I don't care which dot version of 1.10. Just create me a 1.10. I can actually say, in my initial cluster version in Deployment Manager or cluster dash version in gcloud, basically, I can just say =1.10. And it's just going to grab the latest 1.10.
If I want a patch release, I can say 1.10.7, and then it'll always grab the latest one of that with the -gke on the end. Or if I just want the latest, I can just write latest.
MARK: Yeah, so it'll just work it out for you behind the scenes, and it can save you a lot of time, make your scripts a little less brittle, and make things a little bit easier for you.
MELANIE: We like easy.
MELANIE: Cool. Mark, where you going to be?
MARK: I will be at KubeCon in December, but other than that, doing some vacation next month.
[CELEBRATORY ORGAN MUSIC]
MELANIE: I will be at SOCML at the end of November, and that's the main plan.
MARK: Cool, that sounds good.
MELANIE: Yeah. So real quick before we wrap--
MELANIE: --I want to just do a quick little PSA.
[PLAYFUL ORGAN MUSIC]
If you are in the US, please vote.
MELANIE: We're not going to sit here and tell you who to vote for. I've got my own opinions. But-- you can check out my Twitter feed if you want to know my opinions. But yeah, please vote. If you're thinking about it, if you're on the fence, whatever, please definitely vote this midterm election.
MARK: Awesome. Well Melanie, thank you so much for that. And thank you for joining me, yet again, on this week's broadcast.
MELANIE: Thank you, Mark.
MARK: And thank you all for listening. And we'll see you all next week.
Mark Mandel and Melanie Warrick