Chapter 14
Tatyana Yakushev: Data Visualizations and Cloud-Powered AI as Strategic Assets for Next-Gen Analytics
Tatyana Yakushev, principal engineer, Amazon Web Services
Source: Tatyana Yakushev
Tatyana Yakushev is a principal engineer at Amazon Web Services working on Amazon QuickSight. She has been working in the data analytics and data visualization space since 2006. Prior to joining Amazon, she worked as a software engineer or manager at Microsoft, Tableau, and Predixion Software.
Alexander: You are a principal engineer at Amazon Web Services. What is your mission?
Tatyana: Principal engineers at Amazon Web Services are strategic members of product teams, helping to build and manage flexible, architecturally sound systems aligned with business needs. Since data analytics is near and dear to my heart, my mission is to make Amazon QuickSight the preferred service for our customers.
Alexander: How should business models evolve to survive and thrive in an increasingly digital world?
Tatyana: To discuss how business models should evolve to thrive in an increasingly digital world, first we need to talk about the shift to the cloud and the changes it brings. Why are organizations moving their IT infrastructure to the cloud, and why are they doing it so quickly?
- The first reason is cost. With the cloud, you don't have to lay out the capital up front for the servers and the data centers, and you instead get to pay for it as you consume it as a variable expense. If you've ever had to provision infrastructure, you know you could decide to provision on the low side. Then, if it turns out you don't have enough, you create a customer outage; most people don't choose this option. You could instead provision for the peak, but you rarely sit at the peak for long. In the cloud you just provision what you need — if it turns out you need less, you give it back to us and stop paying for it. That variable expense is lower than what virtually every company can do on its own.
- While cost is often a conversation starter, the number-one reason that enterprises and governments are moving to the cloud is the agility and speed with which they can change their customer experience. If you look at most companies' on-premises infrastructure, to get a server typically takes 10 to 12 weeks (sometimes longer), and then you have to build all this surrounding infrastructure software, like compute and storage and database and analytics and machine learning. In the cloud, you can provision thousands of servers in minutes and access a variety of services to get from an idea to implementation several orders of magnitude faster.
What helps organizations thrive in an increasingly digital world? What separates successful organizations from the rest? The biggest differences often come down to a few key factors.
- First, the senior leadership team needs to be aligned and truly committed to the changes. They need to be setting clear direction and expectations for the rest of the organization to get everyone on the same page and working toward the same thing. It's easy for others to do nothing or block things if the leadership team isn't making the transition a priority and building a culture for change.
- Then, the most successful organizations start with an aggressive top-down goal that forces the organization to transition faster than it would organically.
- Third, it's really important that organizations are trained on the technologies and comfortable with the concepts as part of the whole process.
- Last, sometimes organizations can get paralyzed if they can't figure out how to transform every last workload. There is no need to boil the ocean. It is useful to conduct a portfolio analysis, assess each application, and build a plan for what to transform in the short term, medium term, and last. This helps organizations get the benefits for many of their applications much more quickly, and it really helps inform how they move the rest.
Alexander: How can technology shift the roles and responsibilities of the workforce?
Tatyana: Recent technological advances are transforming the way we work, play, and interact with others. These technological advances can create enormous economic and other benefits but can also lead to significant changes for workers.
These technological advances allowed organizations to pivot in the face of the COVID-19 pandemic by providing services remotely and having part of the workforce working from home.
Technological advances allow us to automate routine work and help people focus on more creative tasks. According to a recent McKinsey report,1 “Nearly all occupations will be affected by automation, but only about 5 percent of occupations could be fully automated by currently demonstrated technologies. Many more occupations have portions of their constituent activities that are automatable: we find that about 30 percent of the activities in 60 percent of all occupations could be automated. This means that most workers — from welders to mortgage brokers to CEOs — will work alongside rapidly evolving machines. The nature of these occupations will likely change as a result.”
Automation will increase demand for people with advanced technological skills. Social, emotional, and higher cognitive skills and complex information processing will also see growing demand.
Alexander: Which technology or digital capabilities are essential for a digital strategy?
Tatyana: Artificial intelligence (AI). Even though we have learned to use or even depend on assistant technologies such as Alexa or Siri, in the grand scheme of things, most organizations are still in the very early stages of adopting artificial intelligence. AI will continue to transform every single business.
Alexander: Why do data visualizations have such a strong impact on our decisions?
Tatyana: Good data visualizations show the most important information needed to achieve one or more objectives. This key information is consolidated on a single computer screen so it can be monitored and understood at a glance. Having the information presented in a graphical form allows us to leverage the power of the visual perception, focus on key elements first, notice trends or outliers, and explore further to get the full picture.
Alexander: Why is data storytelling the essential data science skill that everyone needs?
Tatyana: Data analytics can be divided into two broad areas: exploratory analysis and explanatory analysis.
- Exploratory analysis is what you do to understand the data and figure out what might be noteworthy or interesting to highlight to others.
- Explanatory analysis is what you do when you have something specific that you want to communicate to somebody specific. This is where storytelling skills become critical.
Since almost every data scientist needs to communicate their findings to other members of the organization on a regular basis, storytelling becomes an essential skill for every data scientist to master. While many data scientists can build an OK story, it takes years of practice to become a pro in drawing your audience's attention to the things you want them to see, providing the right level of details to answer most common questions without overwhelming the audience, and building a memorable story that others will remember after leaving the meeting and when making a decision.
Alexander: How can everyone learn to communicate better with data?
Tatyana: Step 1 is building data-driven culture within the organization so that decisions are made by looking at the data instead of gut feeling. Step 2 is practicing and learning from others.
Two of my favorite books on the topic of data visualizations are
- Information Dashboard Design: Displaying Data for At-a-Glance Monitoring by Stephen Few
- Storytelling with Data: A Data Visualization Guide for Business Professionals by Cole Nussbaumer Knaflic
Alexander: What basic guidelines and patterns should always be considered for a visual expressive dashboard?
Tatyana: Many data scientists and data analysts focus on the technical aspects of the problem when building a dashboard, such as getting access to the data, figuring out how to automate data transformation pipelines, and figuring out how different metrics are computed. I want more people to think like a UX designer before they start to do any work.
Interview your audience to understand their needs. A great way to do it is to ask them to teach you to do their job as if you were a new member on the team. Listen to understand what information is important for them to take an action or make a decision. Understand where and when they look at the information, what other systems they use, and how they collaborate with other people within or outside of the organization.
Alexander: Companies usually have plenty of legacy dashboards, the message of which cannot be seen at first glance. What would a workshop look like to improve visual data communication?
Tatyana: When helping others build and improve existing dashboards, I like to focus on eight areas:
- Content organization within a dashboard
- Picking the right chart
- Proper use of colors
- Techniques for highlighting what's important
- How and when to use captions, labels, and description to help people understand the content
- Providing necessary context to help compare results to target, another time period, or results in another region
- Techniques to reduce chart junk, remove unnecessary or repetitive information, and increase data-to-ink ratio
- Adding interactivity to dashboards
Before and after examples help people understand and remember the content.
Alexander: Whom should you trust with helping you monitor the changes you're making and testing your dashboard prototype to see if it meets expectations?
Tatyana: Sometimes people forget that dashboards are software applications that benefit from well-established software validation and deployment processes. It might be OK to quickly build a dashboard and share it with the team if everybody is very familiar with the data, the scope of work is very clear, and the dashboard is used by only a handful of users. In other cases, you can adopt processes used by big technology companies. These include the following:
- Automated tests to make sure that data pipelines are functioning correctly and the data is accurate and fresh
- Use of version control systems to track and approve changes and to roll back to the previous version if anything goes wrong
- Early validation by a subset of users to catch problems before they affect everybody
- Usability research to identity if the dashboard meets expectations and how it can be improved. Make sure that this research is done by observing a diverse set of users, including people with different roles, different experiences and backgrounds, different geographic locations, color vision deficiency, or other visual impairments.
Alexander: Why should companies set up an Analytics Center of Excellence as part of their digital strategy? What are some strategies for setting this up?
Tatyana: Becoming a data-driven organization takes time and the efforts of many people. Selecting the right analytics tool and building the necessary dashboards is only a small part of the journey. An Analytics Center of Excellence (ACoE) can help streamline analytics efforts across the organization and steer all of your efforts in the right direction. The ACoE can be responsible for these tasks:
- Setting up your organization's analytics vision. How do you prioritize investments in different areas? How do you measure the business impact?
- Building a technology blueprint. What data, tools, and capabilities will help your organization remain ahead of the competition in the future?
- Establishing standards and best practices to be used across the organization. Besides helping your organization complete analytical projects better and faster, it can help you stay compliant with applicable laws and regulations.
- Managing programs and controlling funding. Should your organization pay to acquire third-party data? How much should you be spending on licenses, salaries, and other expenses?
- Skills development. The analytics area is continuing to grow very quickly. This means that people working on analytical projects need to continuously learn about new tools and techniques and will greatly benefit from knowledge sharing.
When establishing an ACoE, it helps to
- define clear areas of focus,
- secure funding or headcount and establish partnership with other parts of the organization that benefit from the ACoE's work the most,
- build a balanced team with regard to areas of expertise and skills, and
- define demonstrable success criteria and celebrate successes.
Alexander: The Fourth Industrial Revolution is inseparably tied to the vast amounts of data needed to train artificial intelligence. How has this revolution impacted your work at Amazon Web Services?
Tatyana: With the rise in compute power and proliferation of data, machine learning has moved from the periphery to become a core part of businesses and organizations across industries. Gartner forecasts that AI-derived business values are projected to reach $3.9 trillion in 2022.2
AWS has tens of thousands of machine learning (ML) customers who are using ML to drive efficiencies, create new revenue streams, and innovate on behalf of their customers. These companies span many industries, including healthcare and life sciences, finance, technology, retail, media, and entertainment, as well as the public sector.
Alexander: Which AI use cases would you see as essential to contributing to any organization’s digital strategy?
Tatyana: Good machine learning use cases solve real problems for our customers' businesses and can help to gain support and adoption at the organizational level. To assist customers in choosing the right use case, we have prioritized seven horizontal machine learning use cases, which are applicable across a number of industries.
- Personalize Customer Recommendations Amazon Personalize can help applications and websites tailor content to a user's behavior, history, and preferences, boosting engagement and satisfaction. For example:
- A video streaming website can help users discover additional shows that they may be interested in by providing recommendations on the home screen based on past viewing habits and demographics.
- A retailer can recommend items similar to a selected item on a detail page.
- Discover Accurate Information Faster with Cognitive Search AWS makes it easy to combine previously siloed data sources into a central location and use natural language questions instead of simple keywords to get the answers faster and more accurately from unstructured content. Internally, this allows organizations to improve employee productivity, accelerate research, and make faster and better decisions. Externally, organizations can delight their customers with a superior experience and easy access to the information they need.
- Forecast Faster and More Accurately Amazon Forecast is a fully managed service that uses machine learning to combine time-series data with additional variables to build forecasts. It requires no machine learning experience to get started. You only need to provide historical data, plus any additional data that you believe may impact your forecasts.
- Analyze Rich Media Assets and Discover New Insights The AWS Media2Cloud solution makes it easy to import the media content, such as video, audio, and text, into the cloud and then rapidly analyze what the content contains (objects, spoken audio, activities, and so on). Based on these insights customers can build their business use cases, such as content search (internal and external), language detection, audio and text translation, ad classification and targeting, subtitling and localization, content moderation, and so on.
- Add Intelligence to the Contact Center AWS offers solutions that add intelligence to your contact center tailored specifically to your business needs, whether you are building a contact center from scratch or integrating services into existing partner contact centers.
- Identify Fraudulent Online Activities Amazon Fraud Detector is a fully managed service that makes it easy to identify potentially fraudulent online activities such as online payment fraud and the creation of fake accounts.
- Automate Data Extraction and Analysis Organizations have long struggled to process documents efficiently to make them easy to search and access. With Amazon Textract and Amazon Comprehend, AWS helps customers build and maintain a smart index of their document stores. Amazon Textract and Amazon Comprehend use automation and natural language processing to extract and classify documents, and to produce insights that are meaningful to the business.
Alexander: Because so much data is needed to train AI algorithms, how can organizations and companies stay ahead of legal, regulatory, and ethical issues associated with collecting and applying data?
Tatyana: AWS continues to talk to customers, researchers, academics, policymakers, and others to understand how to best balance the benefits of AI technologies with the potential risks. One of the goals of these discussions is to create guidelines for ethical use of AI technologies. The other goal is to encourage policymakers to consider these guidelines as potential legislation and rules in the United States and other countries.
Alexander: Is there a way for citizens to own their data instead of organizations and companies owning fractions of it?
Tatyana: Many countries have laws and regulations to protect citizens' fundamental right to privacy and the protection of personal data, such as General Data Protection Regulation (GDPR) in the European Union. Many requirements under these laws and regulations focus on ensuring effective control and protection of personal data. AWS offers services and resources to help you comply with regulatory and compliance requirements that may apply to your activities. These include the following:
- Encryption of personal data
- Ability to ensure the ongoing confidentiality, integrity, availability, and resilience of processing systems and services
- Ability to restore the availability of and access to personal data in a timely manner in the event of a physical or technical incident
- Processes for regularly testing, assessing, and evaluating the effectiveness of technical and organizational measures for ensuring the security of processing
Alexander: What types of professional roles will we see evolve alongside the development and increasing use of AI across the industries?
Tatyana: The job market of the future will be characterized by human-AI cooperation rather than competition. With increased use of AI, people will be able to do their work better and faster. AI will allow people to spend less time on mundane tasks, such as finding the right information, and more time on creative tasks and human-to-human interaction. For example, a doctor might spend less time entering medical information and more time talking with a patient to make a data-driven decision regarding the best course of treatment. Employees who previously lifted and stacked objects can become robot operators, monitoring the automated arms and resolving issues such as an interruption in the flow of objects.
We will continue to see increasing demand for data analysts and cybersecurity specialists.
Alexander: How do you develop your team's AI skills?
Tatyana: We've been using machine learning across Amazon for more than 20 years and have thousands of engineers focused on machine learning across the company. To help advance our team's AI knowledge, we run a machine learning university (MLU) program. Most of the same courses used to train Amazon employees are available online at no charge3 (you only pay for the services you use in labs and exams during your training).
Alexander: What skills will managers need to develop to be able to work with AI?
Tatyana: Technology companies and departments looking to adopt AI will need to learn how to plan and run their software development programs. AI solutions often require a higher number of experiments and iterations relative to non-AI solutions. This happens because results of training AI models are difficult to predict ahead of time. Managers will need to learn how to set the right goals for their teams, track progress, and help teams focus on the right priorities. Being able to bridge the communication gap between data scientists, engineers, and businesspeople is very beneficial.
Alexander: Thank you, Tatyana. What quick-win advice would you give that is easy for many companies to apply within their digital strategies?
Tatyana: Focus on customer needs and work backward to develop new experiences. Most successful companies got ahead not by competing head-to-head but by rethinking the entire experience.
Alexander: What are your favorite apps, tools, or software that you can't live without?
Tatyana: I like LinkedIn. It helps me stay connected with people working in my field, discover news, articles, books, and courses.
Alexander: Do you have a smart productivity hack or work-related shortcut?
Tatyana: I take time to review my plan for the week, write down what I plan to accomplish, and see how I've done against my goals the previous week. This helps me spend my time more efficiently and not forget something important.
Alexander: What is the best advice you have ever received?
Tatyana: “Take actions to move toward your goal, regardless of whether you know how each one is going to work out or not.” It is not uncommon for people to think that somebody else is more qualified, more capable to achieve something. Women are especially affected by this. Nowadays, I ask myself, “What is the worst thing that can happen?” make peace with it, and move forward.
Key Takeaways
- Most successful companies got ahead not by competing head-to-head but by rethinking the entire experience. Focus on customer needs and work backward to develop new experiences.
- AWS Well-Architected Framework is a great resource for customers to apply best practices in the design, delivery, and maintenance of AWS environments.
- AI will allow people to spend less time on mundane tasks, such as finding the right information, and more time on creative tasks and human-to-human interaction.
Endnotes
- 1 McKinsey Global Institute, “AI, automation, and the future of work: Ten things to solve for,” June 1, 2018 (www.mckinsey.com/featured-insights/future-of-work/ai-automation-and-the-future-of-work-ten-things-to-solve-for).
- 2 Gartner, “Gartner Says Global Artificial Intelligence Business Value to Reach $1.2 Trillion in 2018,” April 25, 2018 (www.gartner.com/en/newsroom/press-releases/2018-04-25-gartner-says-global-artificial-intelligence-business-value-to-reach-1-point-2-trillion-in-2018).
- 3 ML training website:
aws.training/machinelearning.