TLP307: How to Transition from a ‘Knower’ Mindset to a ‘Learner’ Mindset

Joe Schurman teaches from his deep experience in the software, machine learning, AI, and processes that organizations need today as they transition to data-driven technology companies. He names some of the cloud services and tech tools he uses to lead clients to start with a user case, break it into stories,  build a team led by the solution owner, assign the stories to developers to build, and iterate product demos until the Minimum Loved Project (MLP) is achieved. Joe offers observations on investing the “right” amount of time in projects, and wisdom on developing a learner mindset.   Key Takeaways [2:06] Joe Schurman is a 2nd-degree black belt in Kung Fu. He once judged a competition in Las Vegas. He has four children; two daughters and two sons. [2:57] Joe is an expert on the fringes of what we can do with computing technology. What we can do changes every day. In the past couple of years, from an AI perspective, with data and automation, it’s taken leaps and bounds. [4:30] We’re still pretty far away from general AI, despite Sophia, an AI robot that was granted Saudi Arabian citizenship in 2017. Today’s AI depends on the programming we give a machine and its interpretation and output. Joe’s focus is narrow or weak AI. His business colleagues call it magic. Computer vision is an area he loves. [5:45] Joe uses a lab environment across Google Cloud Platform, Microsoft Azure, and Amazon Web Services. The capabilities that have come up in the last year are “just insane” with what you can do with computer vision and building libraries of what the machine can see. [6:06] Joe loved seeing a computer vision capability demonstration at AWS re:Invent of tracking every NFL player on the field and predicting injuries and other types of output and insights in real-time. The machine used narrow AI to access a library seeded with “a ton” of data to interpret the action. [7:15] What you can do with this technology comes down to the data that you feed the engine. Think about the amounts of data that organizations have to sift through to generate reflective or predictive insights. Auto machine learning helps organize the data into useful information such as anomaly detection in software engineering. The data can also come from tools like GitHub and Jira. [8:25] Joe did a fun computer vision project on UAPs for the History Channel, working with some of the nation’s top military leaders, building a library of video and audio data to be able to detect unidentified aerial phenomena that were not supposed to be entering our airspace, and curating that library. [10:06] AI started with the idea of speeding up processes, such as getting an app to market faster or gathering insights quicker to make business decisions more timely. [11:28] AI can enhance human performance. Joe starts by finding people who know how to fail fast; to get a Minimum Viable Product (MVP) out the door. Solutions such as quality engineering automation, test automation, and monitoring services for DevOps detect bugs and performance issues quickly and ensure that the quality of the team is sound.[12:47] Joe notes the importance of individuals performing, contributing to, and collaborating as a team. Set your organization and standards governance up first. Look for a platform of technology to leverage that enables you to build and tinker. Finding the latest and greatest tool is no good unless it provides the right level of collaboration with their platform and connection to different processes. [14:53] When introducing ML to an organization, start with discovery, to understand the culture and talent within the organization. How are they communicating today? Joe sees the biggest gap between data scientists and data engineers. Projects tend to fail without collaboration, regardless of the tech. If the data scientists don’t understand the domain, then the platform is irrelevant,[17:28] Joe stresses the need for a methodology in place to make any of these aspirations work for your organization. After discovery, there’s an align phase. Focus on the outcome and the use case. The solution owner is crucial. The solution owner leads the technology team and brings them together around the client’s outcome to develop that use case.[18:12] If you can’t take an actual use case and break it down into bite-sized chunks or user stories, then the project will never be on the right track. Start with the use case to mitigate risks. Break the use case into user stories. Match the user stories with the number of engineers that can develop a number of user stories within a given time frame. [18:38] Those user stories given to the engineers are deducted into Story Points, in the Agile Process of engineering software. Price Waterhouse Coopers (PcW) has taken it to the next level, being able to do Engineering as a Service, being able to do it at scale, and being able to pivot quickly.[18:58] Joe explains what can happen if you have a great idea, take three to six months to break down the use case, and fill all the requirements, but hand it off to the Dev team that has no idea what the use case is: you get irrelevant software that doesn’t tie back to the outcome! [19:22] Keep the solution team engaged in building the bridge between the subject matter expert stakeholders and the engineers. Every two weeks, demonstrate the iteration or program increment you have built. Does it match the outcome? Does it provide any relevance? Then take the feedback and figure out what happened in that iteration. Fix errors. You will build a product that has value to launch. [20:45] Communicate a lot, so all the people are on the same page! When you have stovepiped organizations where the departments don’t talk to one another, you waste time, effort, and money building a product no one will use. One of Joe’s colleagues, José Reyes, uses the term Minimum Lovable Project (MLP), where people rally around the outcome, not just the tech. [22:33] What skills and knowledge will the leaders of PwC need to endure for the next five years? Joe says first are character and attitude; people that have a hunger to build something, with a fail-fast mentality, and that are excited to learn constantly, that read every day and learn new technology. [24:27] Then know the tools. Documents exist on the internet for every solution and there is access to services like GitHub to download projects and starter templates without being an expert but just reading the README file and installing the base-level template, learning as you go, and as you tinker. That’s way more valuable than coming in as a book-smart expert in a specific product or technology. [24:57] When it comes to tooling, there are products like the Atlassian platform with Confluence and Jira. For an AI stack, Joe typically works with AWS, GPC, and Microsoft, more so on the Amazon side with AWS AI tools, like Rekognition, Glue DataBrew, Redshift ML, Comprehend, and more. Amazon, Microsoft, and Google produce so much documentation and certification to get you up to speed. [26:30] Judgment, wisdom, and character will not be replaced by AI anytime soon. There’s still room for philosophy in leadership. There are tools and technologies to speed up the processes, but not the individuals. There are no general AI solutions out yet to replace a pod of application developers, designers, and solution owners to execute a successful MVP or MLP out the door for a client. [27:55] Advice to CEOs: Be patient and understanding. Be willing to fail fast. Support tinkering and R&D, even if the project doesn't work out. Organizations are generally realizing that today they need to be data-driven, technology companies but there is still hesitance over the risk that needs to be taken. [30:03] Why would an insurance company or other traditional company need R&D? Look at Loonshots, by Safi Bahcall for some ideas about R&D. [30:56] Joe shares how he got to this point in his career. He wanted to play baseball but started at Compaq (now HP) when he was 18, writing scripts in Unix and other environments. Just being able to make certain changes to help clients get products faster and seeing the quick response from the outcomes felt like a home run to him! [31:49] Years later, Joe went on his own, with a vision to create telehealth before telemedicine was a thing, using Skype for Business and Microsoft Lync, enabling an API for that. Seeing people connect through a technology he had built, replaced the need to be a baseball star! Joe is grateful for the break he got at a young age and enjoys his work. [33:22] When Joe first started, he was trying to be the smartest person in the room, seeing the instant gratification of making code snippets that tested successfully. Eventually just building the app wasn’t enough for him. He got the dopamine hit from seeing users interacting with his code and seeing its value. [34:58] Joe’s mentors include many people he worked with. X. D. Wang at Microsoft Research inspired him to tinker, build, and focus on the short-run more than the long-run. Randeep Sing Pal at Microsoft Unified Communications was another great mentor. Also Steve Justice and Chris Mellon, in terms of character and collaboration. Joe shares how they mentored him. [37:23] Jan says something we forget about technology is that there are a lot of failures and attempts before the success hits. We have to be mindful of that as leaders to give people time and space to do really creative, cool things. [38:01] Joe appreciates the opportunity to discuss these things. Joe spent a lot of his career building software solutions that were way ahead of their time. It’s frustrating to see telemedicine so successful now, but not when he attempted it. He had to learn to let go. It’s not just about releasing bleeding-edge tech; you’ve got to find some value associated with it to resonate with the end-user. [39:31] Always think about the outcome and understand your audience first. And then be able to supplement the back end of that with bleeding-edge technology, development, tinkering, failing fast, and all the things that go with software engineering. Also, be humble! Get perspective from outside your bubble to build a better solution and be a better person. [40:49] WHenever you’re setting out to build anything, start with a press release! Write a story of what it would look like if it were released today. Then just work back from there!   Quotable Quotes “There are so many new and cool technologies and innovations that are coming out at the speed of thought, which are pretty fascinating.” “I’ve been in real cloud engineering for about a decade, and from an AI perspective, with data and automation, over the past five to 10 years, in terms of running on a cloud environment, and it’s just taken leaps and bounds.” “You’ve got to be able to connect that [data] environment to a use case or an outcome. If you can’t do that and you can’t enable a data scientist to understand the domain, then the data platform is irrelevant. I see a lot of performance issues occur because of that disconnect.” “If you can’t take an actual use case and break it down into bite-sized chunks or user stories, then the project will never be on the right track.” “In this industry, you’re constantly learning; constantly reading. I’m reading every day and learning about new technology every day and how to apply it and how to tinker with it. I need people on the team … that have that ability or that hunger to tinker and learn.” “Transitioning from a ‘knower’ mindset to a ‘learner’ mindset was the biggest shift for me.” “Always think about the outcome and understand your audience first. And then be able to supplement the back end of that with bleeding-edge technology, development, tinkering, failing fast, and all the things that go with software engineering.”   Resources Mentioned