In this episode Frank and Andy celebrate reaching two hundred thousand downloads, the launch of the Data Channel, also known as Project Ring gate, and talk about the next thousand days.Linkshttps://www.datachannel.tv/homeuse code Launch497 to get 60% off every month forever!

In this episode Frank and Andy speak with Alex Castro on how data can guide project managers to successful outcomes.

In this episode Frank and Andy speak with Alex Murrey, the Head of North American Operations for TangoTeams, a startup that’s redefining the software development outsourcing industry. The software development industry is evolving rapidly and is increasingly becoming data driven, which is why I know you will want to hear what he has to say.

In this episode, Frank and Andy speak to Christopher P. Willis about Using AI to Create a Better User Experience with Better Content.Certainly, if you have ever you read instructions or product documentation that left you annoyed and confused, then you can appreciate the work he does with Acrolinx.Audible  Audible is a sponsor! Click this link and score a free audio book on us! If you subscribe, you help out the podcast. Everyone wins!Transcript 00:00:00 BAILey Hello and welcome to data driven. 00:00:02 BAILey In this episode Frank and Andy speak with Christopher Willis about how artificial intelligence can help bake brands create congruent content across cultures, languages and writers. 00:00:13 BAILey One quick word of correction. 00:00:15 BAILey Frank made the assumption that CPO was chief product Officer. 00:00:19 BAILey Chris is actually Chief pipeline officer. 00:00:21 BAILey In addition to being chief marketing Officer, Acrolinx currently does not have a chief product officer. 00:00:28 BAILey Frank should know by now what happens when you assume anything. 00:00:32 BAILey I'll have a chat with him later. 00:00:34 BAILey For now, enjoy the show. 00:00:44 Frank Hello and welcome to data driven, the podcast where we explore the emerging fields of data science, machine learning and artificial intelligence. 00:00:52 Frank If you'd like to think of data as the new oil, then you can think of us like Car Talk because we focus on where the rubber meets the virtual road and with me on this epic road trip down the information. 00:01:04 Frank Superhighway as always is Andy Leonard. 00:01:07 Frank How's it going Andy? 00:01:08 Andy Good Frank, how are you doing? 00:01:10 Frank I'm doing great, I'm doing great. 00:01:11 Frank It's a beautiful Tuesday morning here in the DC area. 00:01:14 Frank We're recording this on September 28th and I can't believe it's already October. 00:01:22 Andy Almost gosh, yeah, yeah. 00:01:23 Frank Almost October. 00:01:25 Andy It's it's been beautiful fall weather. 00:01:28 Andy Past few days here in sunny Farmville, VA. 00:01:33 Andy And I'm really enjoying that. 00:01:35 Andy Got a lot of outdoors work done in the past few days and that's always a good thing. 00:01:40 Frank Yeah, we just built the trampoline for the kids and that was a was a lot of fun. 00:01:46 Frank 'cause the. 00:01:46 Frank Instructions were horrible. 00:01:50 Andy Did you get one with that big net around it? 00:01:53 Andy Keep from bouncing off and 'cause otherwise it should come with a coupon for a free cast. 00:01:58 Frank Freecast and free healthcare that'd be funny. 00:02:00 Andy Yes, yes, that's right, yeah. 00:02:03 Frank Yeah, so without further ado I'd like to introduce we have this. 00:02:07 Frank Awesome guest today. 00:02:08 Frank We've been really lucking out on terms of folks coming to us and and suggesting guests for us, which is quite refreshing, actually. 00:02:17 Frank So so today we have with us Christopher Willis, Acrolinx Chief Marketing Officer and Chief Product Officer. 00:02:25 Frank Christopher is an expert in technology, marketing and brand alignment alignment with over 20 years of experience in with some of the world biggest tech names including Perfecto. 00:02:35 Frank Kmag and Cambridge technology group. 00:02:39 Frank And through his work at Acrolinx, he's become a renowned thought leader on the topics of content governance and brand alignment. 00:02:46 Frank He's also an expert on AI and how AI can help. 00:02:50 Frank Big brands can great create congruent content across cultures, language and writers. 00:02:56 Frank Acrolinx creates tools for developing content that feels human, relatable, and compassionate. 00:03:01 Frank It's already used by some of the biggest brands in the technology world today, so welcome to the show, Chris. 00:03:09 Christopher Thank you, I'm excited about your trampoline. 00:03:12 Frank Well, thank you. 00:03:12 Frank Thank you. 00:03:13 Frank You should come on down I. 00:03:14 Frank I think you're on the East Coast somewhere in Boston. 00:03:17 Christopher I am outside of Boston, yes? 00:03:18 Frank Awesome cool cool you never know 'cause sometimes people will put where they used to live on LinkedIn and not update that so. 00:03:26 Christopher Nope, haven't gone anywhere in what a year? 00:03:27 Christopher And a. 00:03:27 Christopher Half if not. 00:03:28 Frank Right, right? 00:03:29 Christopher A lot, not a lot of travel, yeah? 00:03:30 Frank Year and a half in the two week lockdown. 00:03:35 Frank Well, welcome to the show so so. 00:03:38 Frank Tell me about what so, so you're a CMO and a CPO. 00:03:43 Frank That's that's, uh. 00:03:45 Frank That's an interesting mix I I can see how the two are related, but can you explain kind of like what it is you do for acrolinx and maybe a little bit about. 00:03:54 Christopher So I do a bunch of things. 00:03:55 Christopher I I I joined the company to run marketing and marketing has. 00:04:00 Christopher A lot of. 00:04:00 Christopher Reach in this organization because of what we do and who we sell to. 00:04:05 Christopher So I reach into pipeline. 00:04:07 Christopher I reach into the product process on product marketing in there and come from a background where this approach really resonates and makes a lot of sense and the way that we collect and build and use data is very aligned to the way that I've. 00:04:24 Christopher I've built content in the past. 00:04:28 Frank Interesting, interesting. 00:04:29 Frank So the the product at acrolinx it it. 00:04:33 Frank It uses AI to create content. 00:04:36 Frank So, so like what does that do is? 00:04:38 Christopher Different so we are. 00:04:38 Frank It kind of NLP. 00:04:40 Christopher We're improving content, so we're we're about being improving the quality and effectiveness of enterprise. 00:04:47 Christopher Content so the easiest way to think about what we do is everybody that writes everybody that owns a content organization, whether that's in a development group with technical documentation or product manuals, or marketing content enablement content, internal education. 00:05:04 Christopher All these folks have a whiteboard in their office and. 00:05:07 Christopher On that whiteboard are all the components of language, the way that they want to create their content. 00:05:12 Christopher It's the tone of voice. 00:05:13 Christopher It's the clarity level education level of their of their readers. 00:05:17 Christopher It's the amount of compassion, emotion, inclusiveness that they want in their contents. 00:05:22 Christopher The words that they want to use and that they don't want. 00:05:24 Christopher Use, it's all up there on the whiteboard. 00:05:26 Christopher They feel good about it. 00:05:28 Christopher They've defined essentially the voice of their group or their organization. 00:05:32 Christopher The problem with that whiteboard is that it's in their office and nobody can see it, and even if they could, we don't have a writers pool in the world anymore. 00:05:40 Christopher We're all writers when you go to work, a byproduct of your work is. 00:05:45 Christopher Content, and so. 00:05:46 Christopher As a marketer I get my best content from folks that don't touch marketing. 00:05:51 Christopher They're just smart people that can create so they don't. 00:05:55 Christopher They don't care about what's on my whiteboard at all, and. 00:06:00 Christopher When we were, I mean, the last seven or eight years you talked about. 00:06:03 Christopher The potential for the digital shift, and I think everybody been using that as a marketing buzzword like digital shift is coming. 00:06:10 Christopher You got to get ready and I don't know if anybody ever really thought it was coming, but it was a great way to so some fear into our prospects that if you don't modernize, the world is going to change. 00:06:20 Christopher Holy crap March hit last year and the digital shift arrives and now you're only touchpoint with your consumer is through digital content for some period of time and it became really apparent to folks that how you commute. 00:06:36 Christopher Gay matters and then all the things that happened last year from from a social standpoint. 00:06:43 Christopher Language took on a very lead role. 00:06:46 Christopher And how do you as an enterprise ensure that you're communicating in the voice of your audience? 00:06:53 Christopher And that's where Acrolinx comes in. 00:06:54 Christopher We look at terminology. 00:06:56 Christopher We look at. 00:06:56 Christopher It's style guidelines voice guidelines to be able to create this essentially central lexecon of how to communicate his business and then. 00:07:07 Christopher Either your writers in real time use acrolinx in their sidebar and what in whatever authoring tool they're using, whether they're using something like madcap flare or Adobe products or Google Docs or Microsoft or anything in a browser. 00:07:21 Christopher Uhm, you're able to use acrolinx in real time to check there your content. 00:07:25 Christopher Acrolinx checks for all the components that. 00:07:28 Christopher It's learned from your organization to create great content and provide you with the score. 00:07:32 Christopher You can improve over time or through automation. 00:07:35 Christopher So think in terms of continuous process, continuous delivery of content. 00:07:42 Christopher I'm checking content in it's being scored delivered back to me. 00:07:45 Christopher I'm making changes and it's rolling out at the speed of my. 00:07:48 Christopher Development process. 00:07:50 Christopher So at the base of what we're doing, when you think about where we're at, it's it's really about taking content in stream of characters, extracting that content, buying the context of that, identifying. 00:08:03 Christopher Uh, your tokens either at the word level at the sentence. 00:08:06 Christopher Level and then adding in the linguistic data underneath that around morphology and compound analysis to understand what's. 00:08:14 Christopher In the content that we're we're looking at identifying terminology and and variant detection and then laying patterns on top of that, our proprietary secret sauce to be able to provide that. 00:08:26 Christopher That feedback of whether or not your content is correct on character on tone on terminology, and then that feeds back. 00:08:34 Christopher To users in the form of guidance. 00:08:37 Frank Interesting, so it guides the people who are creating the content doesn't necessarily generate the content for them. 00:08:45 Christopher We don't override and we don't right because customers. 00:08:47 Frank Right? 00:08:49 Christopher If you think in terms of who our customers are. 00:08:51 Christopher Our our customers tend to be the largest technology companies in the world, so think top 20 global technology companies almost every single one of them uses acrolinx and. 00:09:02 Christopher A piece of guidance might be useful, it might be on purpose, so as an example, when I write through my system and I write the word software. 00:09:12 Christopher It says Are you sure you didn't mean platform? 00:09:14 Christopher And why does it say that? 00:09:15 Christopher Because in my world, if we're talking about our product, I don't want my employees to call it software, it's it's platform. 00:09:21 Christopher It's an extensible platform with integration pieces and an API. 00:09:24 Christopher I don't want to sell it as software. 00:09:26 Christopher But I might have meant to say software. 00:09:29 Christopher So I don't want to. 00:09:30 Christopher I don't want to enforce that rule. 00:09:32 Christopher I want to provide guidance and if you agree with that guidance, you implement that guidance. 00:09:37 Christopher We have the technology to override that, but in almost every case that doesn't make sense to the customer. 00:09:45 Andy So is the input for the content is it? 00:09:46 Frank What is this? 00:09:46 Frank What this? 00:09:50 Andy Is it spoken or written or all of the above? 00:09:55 Christopher All of the above, so we can take in. 00:09:57 Christopher I mean, there's a number of ways to teach the platform to be your editor. 00:10:04 Christopher One is to pull mass quantities of content. 00:10:07 Christopher Give me all your great content. 00:10:09 Christopher What does it look like? 00:10:10 Christopher Identify what you think is good and we're going to read through that, and the system will read through all that content and start generating guidelines. 00:10:15 Christopher Based on what you believe is great content today. 00:10:18 Christopher Uhm, there's also the ability to just go in and into our interface and set guidelines so you can set a tone of voice you can identify how lively you want your content. 00:10:29 Christopher Today there are challenges to all of those methods because over time you're going to learn more, and that's part of what I've really been aiming to. 00:10:39 Christopher Evolve with the product is. 00:10:42 Christopher Gartner, the analyst firm, has said that 50% of of marketers, people that set the company voice are. I don't think this is the word they use, but I. 00:10:52 Christopher Will use guessing. 00:10:54 Christopher I have a good idea of what my audience wants to hear, so I define my tone of voice. 00:10:58 Christopher I define the words that I'm going to use I I think I know what people want to hear. 00:11:02 Christopher And if I use acronyms. 00:11:05 Christopher Go ahead and I take all that information that I've gathered, and I teach acrolinx to to help create content like that and the output of acrolinx is an Acura link score so you're aiming for 100. 00:11:17 Christopher Most customers are aiming for 80. 00:11:18 Christopher You want to be 80 or better. 00:11:20 Christopher 80 means good, 90 means done, numeric value of of good and done, so no subjectivity. 00:11:26 Christopher It's just it is what it is. 00:11:28 Christopher This is either on my guidance or it's not on my guidance and. 00:11:32 Christopher If I get 100 acrolinx score on a piece of content, well by God, that's going to perform fantastically. It's exactly what I think. 00:11:41 Christopher My audience wants to hear and how I think they want to hear it. 00:11:43 Christopher The important word in. 00:11:45 Christopher There though, is I think. 00:11:47 Christopher I think that. 00:11:49 Christopher Where I want to aim to get to is the ability to create a feedback loop from my audience. 00:11:54 Christopher So think in terms of support...

Hello and Welcome to Data Driven.In this episode, Frank and Andy speak with researcher Matteo Interlandi about project Hummingbird.  Audio file matteo-mixdown.mp3 Transcript 00:00:00 BAILey Hello and welcome to dated driven. 00:00:02 BAILey In this episode, Frank and Andy speak with researcher Matteo Interlandi about project Hummingbird. 00:00:09 BAILey Now on with the show. 00:00:10 Frank Second, hello and welcome to data driven. 00:00:21 Frank The podcast where we explore the emerging fields of data science, machine learning and artificial intelligence. 00:00:27 Frank If you'd like to think of data as the new oil, then you can consider us. 00:00:30 Frank Car Talk because we focus on where the rubber meets the virtual road and with me on this epic Rd. 00:00:36 Frank We're on the information superhighway as oh is Andy Leonard. 00:00:39 Frank How you doing Andy? 00:00:40 Andy I'm well Frank, how are? 00:00:41 Frank You I'm doing alright. We're recording this on Wednesday, September 1st, 2021 and the the. 00:00:51 Frank The the remnants of Hurricane Ida are ripping through the DC area. 00:00:57 Frank Uh, so if, uh, if I suddenly get dropped, that's because we probably lost power. 00:01:03 Frank But I do have the backup generator, the one that the professionals installed and my. 00:01:10 Frank Duct taped together a solar generator so. 00:01:15 Frank I will be offline. 00:01:17 Frank For a short. 00:01:18 Frank Bit and hopefully come back online. 00:01:20 Frank How how you doing, Eddie. 00:01:23 Andy I'm doing alright Frank. Well, we are you know I'm about gosh 250 miles South of UM we didn't get near the near the effects of Hurricane Ida as you did. 00:01:34 Andy We're getting a little bit of rain now. 00:01:36 Andy We've had some wind. 00:01:37 Andy Gusts, but it's been really mild, and if you look on the radar. 00:01:41 Andy Gotta watch it into track and I I do. 00:01:43 Andy I'm a weather weenie and amateur but it it just kind of went around us to the to the West and it actually started the east when it got a little north of us and aimed right for your house. 00:01:54 Andy I was looking outside that's where Frank lived, right? 00:01:56 Andy And look, the eye is coming right for. 00:01:58 Andy Frank what's left? 00:02:00 Frank Well, fortunately we're safe. 00:02:02 Frank There was some kind of flooding in Rockville and the small overnight, and some folks they got up. 00:02:09 Frank No one, nobody died that I'm. 00:02:10 Frank Aware of so. 00:02:11 It it says. 00:02:12 Frank You know we're not. 00:02:13 Frank Custom the floods or hurricanes or tornadoes up here in DC and and we're more used to the human threats of, you know, little things like terrorism and things. 00:02:25 Frank Like that, but. 00:02:26 Andy Yeah yeah, you guys got a little bit more to worry about that than we do here in FarmVille, right? 00:02:32 Andy But you know these days. 00:02:33 Andy Who knows? 00:02:35 Andy The, uh, definitely our thoughts and prayers are with the folks in in Louisiana and Mississippi. 00:02:40 Andy They were hit very hard. 00:02:42 Andy I've got got friends in Georgia, Western Georgia were telling me that. 00:02:47 Andy They they took a beating as well and you know it just it looks horrible I. 00:02:53 Andy I you know, I've I've been in a few of those places after hurricanes have hit as part of like church efforts to help clean up and stabilize and stuff like that. 00:03:04 Andy It looks like I don't know. 00:03:06 Andy They people describe it as like a war. 00:03:09 Andy I've never been in a war so I don't know. 00:03:10 Andy I've seen pictures and. 00:03:13 Andy There's a lot. 00:03:14 Andy It looks like a lot of stuff is blowing over, and that sort of. 00:03:16 Andy Stuff, it's just. 00:03:18 Andy So, and they're talking weeks and weeks before power comes back on. 00:03:22 Frank That's horrible, that's. 00:03:23 Andy Similar places, yeah. 00:03:25 Frank That's that's. 00:03:26 Frank Probably going to be do more damage from for a lot of things. 00:03:30 Andy Were you worried? 00:03:30 But on a. 00:03:30 Frank More positive note, uh, a positive note. 00:03:31 Andy Yes, on a positive note. 00:03:35 Frank Uh, we are. 00:03:37 Frank I am super excited to have a special guest and I say super excited because he's from Microsoft. 00:03:42 Frank He's a senior scientist in Jelt at Microsoft, working on scalable machine learning systems. 00:03:50 Frank Before he was at Microsoft, he was a postdoc scholar at the Computer Science department at UCLA, and this he was doing a lot of interesting stuff there. 00:04:03 Frank He was doing research at Qatar or Qatar. 00:04:05 Frank I'm not sure how to say that exactly, but he has a PhD in computer science. 00:04:11 Frank In university. 00:04:12 Frank Of Modena and or? 00:04:15 Frank I'm going to botch this. 00:04:15 Frank Reggio Emilia. 00:04:17 Frank Welcome to the show, Mateo. 00:04:22 Frank Awesome, so we are really excited to have you here. 00:04:25 Frank We actually booked you a whole month in advance. 00:04:27 Frank I've been looking forward to this. 00:04:29 Frank Yeah, because you're coming by way of some of the folks at the Mlad conference. 00:04:35 Frank And for those who don't know, I'm a I've mentioned this. 00:04:37 Frank Mlad stands for machine learning and data science summit. 00:04:40 Frank It used to be in person I think now it's entirely virtual for the foreseeable future. 00:04:45 Frank Uh, but that why I attended M lads in 2016 summer of 2016 and it was uh, it was life altering like I don't say that. 00:04:55 Frank Lightly so. 00:04:56 Frank So Microsoft does amazing work in the machine learning and data science space. 00:05:02 Frank Very much cutting edge stuff very much I. 00:05:06 Frank I wouldn't say under the radar, but Microsoft does not do a great job putting its own horn, so we're very excited for you to come on Mateo and talk about this little project that you're working on. 00:05:17 Frank And what is the is it have a code name or what? 00:05:20 Frank What is it called? 00:05:22 Matteo Hummingbird should the code name is actually I'm in. 00:05:26 Matteo Don't have any specific internal names for. 00:05:28 Matteo This for this. 00:05:28 Frank OK, what what is GL stand for? 00:05:32 Frank That was my that was my first question. 00:05:33 Frank When I saw your bio. 00:05:35 Matteo Uh is for Gray system lamp and is the after Jim Gray which. 00:05:41 Oh, OK. 00:05:41 Matteo Is putting award yeah? 00:05:45 OK. 00:05:46 Matteo So these are the search lab after this name yeah and use within the Azure data organization. 00:05:49 Oh, interesting. 00:05:53 Frank And uhm, So what? 00:05:56 Frank What what cool stuff does Hummingbird do? 00:06:00 Matteo So, Hummingbird, uh? 00:06:03 Matteo Is a little bit, uh, weird project in the sense that when we started this project we didn't know if it was going to. 00:06:10 Matteo To be a success or not? 00:06:12 Matteo Because what we try to do basically is to uhm translate traditional machine learning models and into neural networks. 00:06:22 Matteo Actually not Internet format into tensor programs such that then we can run over tensor runtime, such as pipers. 00:06:30 Matteo In terms of. 00:06:32 Matteo Uhm, so when we started this project actually idea was hey there is a lot of investment in general pulling into this neural network frameworks and. 00:06:45 Matteo Coming from the Azure data organization, instead, we are more interested in these traditional machine learning methods such as decision trees. 00:06:52 Matteo Linear models were not encoding all those boring traditional algorithms. 00:07:00 Matteo And so we look at this. 00:07:01 Matteo The neural network system and say hey how we can take advantage of all this technology that is built. 00:07:05 Matteo Into this domain so you can run neural. 00:07:08 Matteo Network over CPU. 00:07:10 Matteo Over the GPU, then you can use like fancy compilers to compile to generate the transfer programs. 00:07:16 Matteo All those sort of techniques and we were. 00:07:19 Matteo Kind of struggling. 00:07:20 Matteo To see what we could do with the with this stack and and what we come up with with is this Amber project. 00:07:27 Matteo So we basically take a. 00:07:32 Matteo Traditional machine learning pipelines composed right feature iser and machine learning models. 00:07:37 Matteo After the day trained. 00:07:39 Matteo So first you need to train it using cycle ornamental net or. 00:07:43 Matteo Uhm, uhm, one of those traditional machine learning platforms and then once it is trained we basically convert it into a set of tensor operations in. 00:07:54 Matteo In the current version we use basically PY torch for doing this conversion and then basically you have a pipeline model so you can do whatever you can do with Python. 00:08:03 Matteo Models so you can deploy it in in it into a PY torch. 00:08:08 Matteo Uhm, deployments you can run over CPU ran over the GPU or you can do the torch script if you want to get rid of all the Python dependency and just have a C++ program you can. 00:08:19 Matteo Do all those all those tricks. 00:08:22 Frank Interesting, does it impact accuracy precision? 00:08:26 Frank Does it improve it? 00:08:27 Frank Keep it the same. 00:08:29 Matteo We tried to keep it the same so we are able to keep. 00:08:33 Matteo It The same up to floating point numbers roundings? 00:08:36 Matteo So since we use, you know we use PY torch to run these programs and not like a socket or ornamental net. 00:08:44 Matteo There are some differences in how they do you know, floating point operations. 00:08:48 Matteo So the. 00:08:49 Matteo Accuracy is up to roundings in the Floating Points, which sometimes are actually. 00:08:54 Matteo It can be quite a bit, but most of the time is really small, almost not noticeable. 00:09:00 Frank Interesting, interesting, uhm. 00:09:03 Frank Do you would you know. 00:09:05 Frank If there was like. 00:09:06 Frank A discrepancy, or you Dutch as part of testing? 00:09:09 Matteo It's part of testing. 00:09:10 Frank Right, all software is tested, right Andy? 00:09:11 Matteo So we have we have. 00:09:13 Frank Sometimes intentionally is that the email. 00:09:15 Andy That's right. 00:09:17 Frank And he has a saying where all softwares I I forget exactly what it is. 00:09:21 Frank But what is it? 00:09:23 Andy Yeah, all software is tested, some intentionally. 00:09:27 Frank There you go. 00:09:30 Frank Uhm, so what's the? 00:09:33 Frank What's the real? 00:09:34 Frank What are? 00:09:34 Frank What are the advantages of of of converting kind of a traditional model over to a tensor model? 00:09:41 Frank Is it? 00:09:41 Frank Is it portability? 00:09:42 Frank Is it speed? 00:09:43 Frank You did mention that you can run it on. 00:09:45 Frank You could take advantage of GPU as well as CPU. 00:09:51 Matteo Yes, exactly so you most mostly is related to speed, so you can basically run your socket, learn model on GPU end to end and and this user provides you know a little bit of quite a bit of speed up we for some of our example we even saw like 2 ordinal Magneto speedups. 00:10:11 Matteo For some of the models. 00:10:13 Matteo And uhm, and usually we try to show that. 00:10:18 Matteo If you use GPU. 00:10:19 Matteo Can be much faster, but on CPU we try to be kind of as close as possible scikit learn or the base or the base or diminished model. 00:10:27 Matteo Sometimes we can, sometimes we are a little bit slower. 00:10:31 Matteo Uh, but we. 00:10:32 Matteo We had some really interesting result. 00:10:34 Matteo Like for instance, we did some experiment with some. 00:10:39 Matteo Some folks at the VM and we took some extra boost model and we compiled some training accuracy boost model. 00:10:47 Matteo Uh, using Hummingbird anti VM into some uh, we basically do code generation and we show that the that model that was compiled to Python was even faster than they quoted the C++ implementation that they're having next used, but those CPU and GPU. Yeah, there was kind of OK. What's going on? 00:11:06 Matteo This is not. 00:11:08 Matteo This was not expected. 00:11:08 Frank Wait, did you say it was faster than a C++ implementation? 00:11:11 Matteo Yes, I mean if she used. 00:11:13 Matteo Underneath C++ even scikit learn. 00:11:15 Matteo You know they use like. 00:11:16 Matteo From C++ library and yeah, using TVM for doing the code generation, they are able to do like a operator fusion which you don't normally have for like these traditional models. 00:11:28 Matteo So we told these tricks bigger, basically that are coming from the neural network. 00:11:31 Matteo Famous we were able to get like this. 00:11:34 Matteo These surprising numbers. 00:11:36 Frank Interesting, so that's a real performance boost, and probably if you scale that up into the cloud that probably. 00:11:44 Frank Means a lot of money saving too in terms of on cloud computing things like, I imagine a company like the size of Microsoft would be very interested in getting better results faster with less cloud compute. 00:11:56 Frank You did mention an acronym, I just wanna make sure folks know. 00:11:59 Frank What that is? 00:12:00 Frank Tyvm what is that? 00:12:03 Matteo Uh, I don't know what is exactly for, uh, some tensor maybe? 00:12:08 Frank Andy looks like he knows, but he's on mute. 00:12:10 Andy I don't, yeah I I don't know. 00:12:13 Frank OK, I'm just curious. 00:12:13 Andy I'll go look it up. 00:12:15 Frank There you go. 00:12:16 Andy EVM acronym. 00:12:19 Matteo I think is for tensor virtual machine, but I'm. 00:12:21 Matteo Not sure if this is approach. 00:12:22 Frank That sounds about right. 00:12:23 Frank Tector,...

In this episode, Frank sits down with Himanesh Go elle, a recent graduate of Virginia Commonwealth University, where they discuss the academic curriculum around data analytics and career prospects for graduates in the field. 

In this episode, Frank and Andy chat with Tyler Browder.Tyler Browder is the CEO and Co-Founder of Kubos, the world’s first cloud-based mission control software.Kubos’s “Major Tom” software is a cutting edge mission control platform for low-earth orbit satellites. Data and space. Does it get any cooler than that?

Here's another bonus episode that BAILey has put together.Who's BAILey? Glad you asked. In the intro, she has a thing or two to say.In this session from the Azure Global Data Fest, Veronika Kolesnikova tell us how make your applications interactive with Speech Services.Original YouTube video: https://www.youtube.com/watch?v=b4RSZ7aIKgE

In this talk from the Azure Data Fest held on June 25, 2021, Jon Tupitza explores automated ML in the Azure ML service.

While Frank is on holiday and Andy is occupied elsewhere, I thought it would be a good time to take over the show this weekend and share some special bonus content.The following is the keynote for the Azure Cloud Events conference, wherein Frank talks about the future of AI and the top technologies to watch.Watch on YouTube: https://www.youtube.com/watch?v=dSbMZjrgCFI