Should Your Digital Transformation Partner Be a Tech Guru?

Enterprises across the globe are changing the way they operate in fundamental ways by generating new revenue streams from digital products or reinventing the customer experience and core value propositions.

A digital transformation should be viewed as a strategic effort led by the CEO, not just as a technology modernization project, but as much more than that. It’s about finding new ways to create competitive advantages, delivering experiences over features, disrupting your own current value engines and replacing them with new business models made possible by new technologies like artificial intelligence (AI).  This requires a human-centric approach that balances business goals, technology capabilities and stakeholder and user needs. Since it takes a village to transform a business, an optimal team integrates internal resources with external digital transformation partners that will provide a balanced combination of business, design and technology perspectives. This view is well-aligned with design thinking principles.

Business Perspective

Understanding business goals and objectives, designing sustainable business models, measuring business impact, and determining business viability will require the following:

  • Business Leaders– Executive sponsorship, business leadership and subject matter expertise are unequivocally critical for any major business transformation. CxOs and leaders from business units need to be vested and committed to be catalysts for change.
  • Industry Experts- Internal or external resources with industry and domain expertise can offer fresh perspectives on the competitive landscape, recent trends and market analysis. Business and domain consultants will be key in conducting market research, ROI analysis and to design the new business models, revenues and KPIs for new product and service offerings.

Human and Design Perspective

Grounding business goals and strategies in user needs for new solutions, products and services will be the key goal of the design group. Consider the following people-centric roles in your team:

  • Customer Experience Crafters– Service and user experience designers are the crucial link between the business vision and customer needs. They interview and bond with the customers, understanding their human needs, wants, limitations, and desires. They design storybooks for customer journeys around new product and service offerings. They will make sure that every great big idea is addressing the right problems and that they are delivered as delightful and compelling experiences for customers and users.
  • Changemakers– A business transformation will affect people in your organization in multiple ways: from altering daily routines to fundamentally changing the way people work. Some will manage the transition better than others. Change management professionals can work with your business leaders to understand the needs of your employees and make sure they are properly equipped to manage each stage of the transition.
  • Real users– Real-live customers can be the best digital transformation partners. Organizations committed to human-centric design often include end customers throughout the transformation process. They will be the actual voice of the customer when formulating ideas, evaluating vendors and products, designing solutions and beta testing a product.

Technical Perspective

The tech team will execute the vision and create digital solutions that are innovative and also feasible to implement.

  • Tech Gurus– There can be no digital transformation without digital experts. Technology partners fuse creativity and engineering discipline and focus on execution and delivery. They have experience implementing emerging technologies such as AI, Internet of Things (IoT) and virtual reality (VR) and understand the potential and limitations of each. They inject feasibility into the digital innovation ideas and services and employ agile development best practices to deliver quality products.
  • Insightful Data Whizs– In today’s knowledge economy, insights providers focus on harnessing the power of your data to extract novel observations and gain a new understanding of your customers, your business, your industry or the world. AI is enabling companies to shift their focus from selling unique products to injecting extraordinary wisdom into their offerings. Establishing information goals and a data strategy to support it will play a pivotal role in the results of your business transformation.
Digital Transformation Partners

Figure 1: Digital Transformation Partners’ capabilities aligned with Design Thinking framework.

Who Will Lead the Transformation?

Welcome innovators! Many companies hire external partners to lead the collaboration between the business, design, and technology teams.  They can help people see emerging future trends, balancing the three perspectives and facilitating conversations so that the best insights, ideas and decisions are not dominated by any one point of view.

  • Creative Innovators-Innovation consultants who have worked with leaders in a wide range of industries will be a great asset to introduce new thinking to business problems and to build an innovation culture within the transformation team. They offer great facilitation techniques to foster creativity, spark curiosity, move the teams to push the boundaries on what is feasible and inspire everyone to THINK BIG. Innovation consultants will prove to be great assets to your digital transformation partner network.

Can Innovators Also Be Tech Gurus?

Human problems are often disguised as technology problems. This has led tech gurus to evolve to become innovators.

Wovenware’s deep expertise in building custom AI and software engineering solutions is in high demand by many business leaders looking to innovate in their organizations. When we are contacted by a CIO who wants to implement AI solutions, we take a design driven approach to understand the underlying business problems and customer pain points before undertaking any technical work. Being able to work at the intersection of business, design and technology is key to leading successful strategic projects. Dedicating enough time dive deep into empathizing with people, understanding business impact and designing new sustainable business models, experiences and solutions are critical for any digital transformation.

Selecting Digital Transformation Partners

An effective digital transformation team will likely require the collaboration among many people with different perspectives and skills. Choose your partners with the same standards you set when hiring new employees. They must have the right skill-set and experience, but they must also share common values and fit the business culture. Unless you are set on partnering with one of the top five consulting giants, you will likely be joining forces with multiple companies that have niche specialties. When evaluating digital transformation partners, you should follow these basic guidelines:

  • Prioritize business culture compatibility
    Success is driven by a cohesive team of talented human beings who can work well together and build a trusting relationship. Engage in multiple conversations with your vendor, pay them a visit and get to know their people. Would you permanently hire them? If your gut answer is no, they may not be the right partner for you regardless of the sophisticated experience they may have.
  • Value all related experience
    Digital transformations generally involve innovative business models and emerging technologies. Don’t set out to search for vendors that only have previous experience with specific technologies or business models in your domain. You should equally value related experiences and proven results with other business transformations, understanding your partners will work alongside you to experiment, test and learn in order to break new ground.
  • Always check references
    When checking references, you will find that other business leaders are very empathetic to your situation and the importance of the journey you are about to undertake. You will likely get very honest opinions about the strengths and weaknesses of a vendor, so do not pass on a valuable opportunity. You can validate your gut feeling or identify some red flags to consider.

Wovenware helps businesses navigate the complexity of digital transformation by discovering challenges and insights, identifying strategic opportunities, designing new services and ideas and delivering software and AI solutions. We employ multidisciplinary teams of innovation consultants, industry experts, service and customer experience designers, software engineers and data scientists to drive business transformations.

Digital Transformation Examples Beyond the Mainstream – The Spark to Action For Businesses Everywhere

Digital Transformation is perhaps one of the most trending buzz phrases in business blogs and publications. It is a loosely used descriptor for a wide range of strategic objectives: from tactical projects to cloud applications and novel products and services that disrupt industries. To do justice to its meaning, a digital transformation must fundamentally change an organization’s way of doing business, whether this implies redesigning the customer experience, uprooting and reinventing operational processes or creating new revenue streams with technology. Most modern enterprises will tap into the powerful and ever so appealing capabilities of new and emerging technologies to drive an impactful business transformation. While some major transformations are widely known, like Netflix’s shift from a DVD rental retailer to an AI-powered video streaming service leader, there are plenty of non-mainstream digital transformation examples that can be the spark to action for businesses that are lagging behind in the quest to remain competitive.

Organizations that operate in regulated industries face a set of challenges when it comes to implementing innovative ideas. Information security, privacy regulations and auditory compliance take center stage in every technology project, adding a few complicated but necessary layers to the innovation process. Following are digital transformation examples in the healthcare, banking and insurance industries that should motivate business leaders to think creatively and identify new ways to impact their organizations.

Credit Unions Play to Their Customer Service Strength

Credit Unions are known for offering better customer service and higher interest rates on deposits than banks but have been lagging in digital innovation. They have been facing pressure from millennials to get up to date with technology for some years now, but since the COVID-19 pandemic has forced many physical branches to shut down, a digital transformation in the industry is imminent.

As an example, Affinity Plus Federal Credit Union has gained a unique edge after undergoing a three-year extensive digital transformation. It fundamentally changed the digital and in-person experience and reengineered its operations. Catering to millennials looking for intuitive, digitally led solutions, it implemented an innovative, mobile-first digital banking platform. It also changed its 20-year-old core operating system to provide members and employees a better experience in terms of  day-to-day interactions. Understanding that not all members are technically adept, Affinity Plus also placed significant investment in developing next generation branches, which provide open technology bars, face-to-face interactions and a wide array of new digital tools. Affinity Plus showcases a great digital transformation that is driven to make technology work for people, not the other way around.

Hospitals Improve Administrative Interactions with Patients

Clinical applications of AI, such as computer vision algorithms that detect cancer in X-ray images, have become the “poster child” of the future of healthcare. However, there are many other areas that affect a patient’s experience, including navigating through administrative processes like pre-registration or managing finances.

Boston Children’s Hospital is using AI to transform the patient-provider communication around financial and administrative services. The hospital is automating the process of obtaining managed care authorizations and approvals for different procedures and outpatient visits, freeing up staff to have more meaningful interactions with patients. Beyond that, it is also coming up with better ways to communicate with patients and their families about their out-of-pocket costs as deductibles and co-insurance costs continue to climb. It is doing this by providing self-service options that will enable the hospital to expand its hours of operations beyond the traditional call center window.

Insurance Companies Look Beyond Quantitative Analytics

The insurance industry is no stranger to sophisticated quantitative algorithms and predictive analytics as part of the risk assessment process. Newer advances in machine learning and computer vision are creating new core value propositions for insurance customers. Amica Mutual Insurance is leveraging satellite imagery to do automatic remote inspections of properties, allowing them to respond to customer needs with unprecedented speed and quality of service. With the new AI platform, the Amica team could, for example, detect if a roof is damaged, assess risk and proactively reach out to customers before the structure is severely compromised.

PBM Launches Prescription Home Delivery Application

Pharmacy Benefits Management (PBM) companies administer prescription drug benefits on behalf of health insurers. They manage the nitty gritty intricacies and complexities in claims processing and formulary management. As an innovator, Abarca and Triple-S Management Corporation are on a mission to continuously improve the pharmacy experience for patients. In 2019, they partnered with software and AI consultancy Wovenware, to build Puerto Rico’s first prescription home delivery program. The major product launch coincided with the COVID-19 pandemic and “Triple-S en Casa” (the name of the service) provided the convenience of home delivery and peace of mind for millions of people in Puerto Rico at a time when they need it most.

Approaching Healthcare Data Security in a Different Way

In highly regulated industries, digital innovation is bound to strict security standards. Customer-sensitive information must never be compromised and data privacy must be protected at all costs. Innovations in cybersecurity may not be as appealing as other mainstream digital transformation examples, but they have a profound impact on a company’s bottom line. Data breaches cost companies millions of dollars every year and attacks are increasing in numbers and severity and they are constantly changing. Rules-based software has been guarding patient information for the past decades, but machine learning algorithms may offer new and more effective ways to safeguard data. Johns Hopkins developed an AI application to produce a highly accurate privacy analytics model that reviewed every access point to patient data and detected when an electronic health record (EHR) was potentially exposed to a privacy violation, attack or breach. Employing supervised and unsupervised machine learning techniques and transparent AI methods, Johns Hopkins implemented a predictive privacy analytics platform. As a result, security investigations were reduced from 75 minutes to only five minutes. Further, the false-positive rate dropped from 83 percent to an astounding three percent, indicating that most notifications were real data breaches. With a new approach to compliance analytics Johns Hopkins fundamentally changed the operational efficiency, potentially saving millions of dollars and protecting patients’ trust that when compromised, can be impossible to regain.

Taking Action

To achieve results comparable to the ones in the digital transformation examples described above, most companies partner with external organizations with different value-added services such as innovation consulting, service design, agile software engineering and applied AI solution development. Multidisciplinary teams will be the key to powering your digital transformation and should include business leaders, industry experts, tech gurus, creative innovators as well as actual customers.

Wovenware, Inc. is a technology consulting firm providing end to end services to deliver digital innovation. Contact us if you are looking for a partner to drive your digital business transformation.

Understanding Deep Learning-Based Object Detection Models with Saliency Maps

Increasing amounts of available satellite imagery has led to advances in the development of aerospace applications due to a wealth of information that needs to be analyzed. This has resulted in the growth of deep learning, an effective AI tool for object detection tasks and broad area search in satellite images.

Wovenware’s data science team often works with deep learning models that have applications in broad area searches on satellite imagery. In this blog, I will discuss how saliency maps can be used to visualize what regions or objects our models consider as important features while we train our object detection models. To understand this, I will document a salient map analysis based on our recent work with solar panel detection.

Since the insights obtained from the analysis performed with deep learning models can often have drastic impacts on decision making, public policy, and the lives of the general population, some deep learning applications require accountability and transparency about their decision making process. This means that we need to be able to explain the inclusion of features in our data and justify our model’s predictions. Although the “black box” that constitutes deep learning is still unresolved, there are ways to visually demonstrate what features drive the decisions made by our object detection models. Saliency maps, usually represented as heat maps, help us identify what areas of an image are the most important to the object detector’s final decision, and allow us to asses if the areas are directly related to the object of interest. Saliency can help us visualize biases acquired by the model from the dataset, originating in the annotation process or any other source.

Solar Panel Detection Project

In order to observe what insights can be obtained from our object detection models for satellite imagery, saliency maps have been generated from their features maps. The generated visuals are then used as proxy explanations of our model’s decision making process. We will explain how these visualizations can help prevent bias in our data and further improve the data collection process. The objective is to demonstrate how saliency analysis can improve the way we enhance our object detection models for satellite imagery within the solar panel detection context.

Like any other machine learning project, the first step consisted of a data collection and curation stage that included a thorough annotation process of solar panels. Then, in the second stage, a series of convolutional neural networks were trained with the resulting datasets. Saliency maps were generated as we trained our baseline model. We consequently compared the resulting visuals to the actual detections made by the model. In Figure 1-A , we can see that our model correctly identified solar panels with red bounding boxes. In order to visually understand what drove our model to correctly detect solar panels, we created a saliency map, shown in Figure 1-B, to our input image.

Figure 1: A) Left – predictions over image.

Figure 1: A) Left – predictions over image.

B) Right – saliency map.

B) Right – saliency map.

From the images in Figure 1 we can see that the model correctly recognizes the area containing the solar panels to be important. Despite this, since we can see a rectangular outline around the area that contains the panels, it seems that the model also recognizes the house’s roof. This suggests that rooftops contribute to its final decision on whether a solar panel has been detected. Based on this information, we can only imagine that our model would struggle in areas where solar panels are not always installed on rooftops. Therefore, we tested our model to see if it would find panels on a solar farm (Figure 2-A) but it was unable to find any in the area. This suggests that our model underperforms when it comes to detecting solar panels in solar farms as it was only trained with panels on rooftops. Let’s see what insights can be obtained from the saliency map generated for these predictions.

Figure 2: A) Left – predictions over image.

Figure 2: A) Left – predictions over image.

 B) Right – saliency map.

B) Right – saliency map.

We can see from Figure 2-B that our model doesn’t know what areas are more likely to contain solar panels in the farm. In order to eliminate the rooftop bias and improve our model’s ability to detect solar panels on solar farms, we needed to go back to our data collection and curation stage and find panels located in solar farms. We then trained and tested our new model against the same images.

Figure 3: A) Left – predictions over image.

Figure 3: A) Left – predictions over image.

B) Right – saliency map

B) Right – saliency map

The new model seems to be an improvement as it identified, in Figure 3-A, an additional panel that went undetected by our previous model in Figure 1-A.  We still don’t know if it relies on the roof to make the detections. The saliency map shown in Figure 3-B doesn’t show the roof’s outline seen with our previous model in Figure 1-B, further supporting the notion that we have reduced the bias that favored panels on rooftops. Now, we can expect our model to be able to find solar panels on solar farms.

Figure 4: A) Left – predictions over image. 

Figure 4: A) Left – predictions over image.

B) Right – saliency map

B) Right – saliency map 

We can see in Figure 4-A that, even though our new model did not find all the panels in the solar farm, it did find most of them. In this sense, it greatly improved when compared to the previous one, which couldn’t find a single panel on the farm. The saliency map in Figure 4-B supports the premise that our new model is capable of finding solar panels of various types and in different regions.

In the end, saliency maps can indeed provide us with important insights about our models and their decision making process. During this analysis, we first noticed that, in order to improve our models, we needed to acquire more solar panel data, specifically ones that where located in solar farms. The visual explanations generated by the saliency maps guided the data collection process, saving us time and minimizing costs, since acquiring satellite image data can be a time consuming and expensive endeavor.

Where was AI During the COVID-19 Pandemic? And what About the Next One?

There has been a debate going on about how much AI has contributed to combatting COVID-19, in terms of prevention, treatment or the search for a vaccine. While the work being conducted by biopharma researchers and drug manufacturers has taken center stage – and rightfully so – AI has quietly worked behind the scenes, most notably in finding and predicting the pandemic in the first place.

For example, Lawrence Berkeley National Lab in California is using an AI program,, to sort through volumes of available research for any relevant data that could be helpful for treating patients and developing a vaccine. Researchers at Flinders University used AI to analyze the virus and how it infects cells, so that a vaccine could block that process. And, Northwestern University is using an AI program to rate and prioritize research in order to determine which programs should receive funding and fast-track their status for vaccine development.

It’s not just smart apps working on the front lines, but also chatbots, which are providing vital information to the public without straining staff resources. Healthcare facilities are turning to chatbots or conversational AI, which uses natural language processing to simulate human conversation. Providence St. Joseph Health in Washington State implemented a Coronavirus Assessment Tool online to educate the public about the potential symptoms of the virus and help people figure out if they should be seen by a healthcare professional. In its first day alone, more than 500,000 people used the chatbot. It not only gave people critical and immediate access to information, but it also freed up medical staff to focus on treating their patients.

AI is certainly not a panacea to magically solve every problem. Its purpose is to augment rather than replace human intelligence, and it depends on humans to provide the right conditions in order for it to be effective. First of all, AI needs to have a huge volume of good data. Since AI learns based on the data it is provided, the more reliable data it has, the more accurate it can be in identifying patterns that can lead to insights and innovation.

Additionally, this data must correlate to the problem that the AI program is trying to solve. Unfortunately, much of the data that AI programs need to address COVID-19 issues is just not available yet. Without the availability of this data, it is difficult, for example, for AI to find potential matches between symptoms and drugs for treatment.

One of the key factors contributing to the lack of data is that the information is unfolding in real time, and everyone is scrambling to come up with vaccines and treatments in the midst of the pandemic; it’s like trying to build a plane while flying it. Another roadblock is that much of the data that is needed is siloed within different labs and offices across countries, although efforts are underway to centralize this data.

Putting AI into Action in the Race to Treat COVID-19

Despite the challenges, AI is playing an important supporting role in addressing COVID-19, but it requires a collaborative and standardized approach that goes beyond private and public interests to succeed. Here are four key steps that can be taken to ensure that AI lives up to its full potential in this pandemic and future crises:

  • Carve out the appropriate role for AI and plan ahead. Any AI project requires a methodical approach despite the urgency. It’s important to leverage the strengths of AI, identify the problems you want it to solve, the type data that is required, and how that data will be collected.
  • Create an international AI taskforce. This should include representatives from as many countries as possible, so that for this pandemic or other crises that emerge, we can be prepared to address them collectively. We need to be able to identify and collect the right data quickly, and have consistent processes and procedures in place to do so.
  • Establish an international database. It should include all the data from around the world, which becomes available to everyone, in much the same way that the gene sequencing of the virus was shared globally. This is a time for collaboration, not competition, for the greater good. With the COVID-19 pandemic, we can see why diverse data from everywhere is critical. For example, Sweden took a very different approach than other countries regarding social distancing and quarantining. Their data will provide critical information that may not be available elsewhere.
  • Ease up a bit on privacy. There has to be privacy tradeoffs when it comes to collecting data. While privacy is certainly important, when it comes to combating a pandemic that can mean life or death for hundreds of thousands of people worldwide, it makes sense to ease up on some privacy concerns for the greater good, while working to find a good balance between privacy and safety.

As we’ve heard throughout the COVID-19 pandemic, data needs to drive the bus when it comes to addressing the many problems that have emerged. While AI certainly does not hold the cure for COVID-19, it can play a critical role in helping us turn data-driven insights into action. Yet, it requires global collaboration among businesses and governments, an understanding of its possibilities and limitations, and bright minds to guide its proper use today and into the future.

Guiding Principles for Data Science Project Managers

The number of companies around the globe that are adopting artificial intelligence (AI) multiplies every year. Given this trend, data scientists are in high demand and specializations like natural language processing (NLP) and deep learning are becoming increasingly important. Despite the shortage of AI talent, companies cannot ignore the role of soft skills, such as storytelling and effective communication, which is critical to effective execution of data science projects. Data science project management must be an effort to bridge the gap between the data science team and other business units to bring out these essential soft skills and foster greater collaboration enterprise-wide.

While the data science manager will play an increasingly larger role in bringing data science and business units together, currently, job postings for data science managers are a very small fraction of postings for data scientists. Since it is an emerging career, the minimum qualifications vary greatly – from PhD degrees in data science with 6-plus years of experience (targeting a very limited pool of applicants) to traditional managers with strong communication skills and knowledge of project management principles and concepts (which is almost equal to “no experience necessary”). Wovenware has taken the approach of training traditional project managers, scrum masters and technical leaders to be able to lead data science projects. As part of our company strategy, we are making sure all our leads are AI-literate, have a general understanding of AI business use cases and the experimental process. This is the first step, but we are extending the program to provide opportunities for up-skill in basic understanding of statistics, data analytics, and storytelling. We will continue to develop the program as a new generation of specialized and formally trained data science project managers emerges.

The problem is that when it comes to data science project managers, the role continues to change and evolve. Needless to say, there are no nationally accepted degrees or professional certifications (yet) to give leaders formal training and a competitive edge in the market. The following guiding principles are a good starting point for leads that are transitioning to data science management.

A New Set of Guiding Principles for Data Science Project Managers

Traditional project managers are almost hard-wired to do whatever it takes to deliver results on time and on budget. To drive innovation in an organization and lead data science teams, managers need to follow a very different set of guiding principles. Famous author of The Mythical Man Month, Fred Brooks, accurately describes it this way: “A scientist builds in order to learn. An engineer learns in order to build.” The stark contrast is self-evident when comparing the two management approaches side by side.

Data Science Project Management

Figure 1: Traditional Project Management vs AI-Driven Innovation

Building vs. learning– Projects will no longer be driven by deterministic goals of building an automated machine but will be driven by a broader vision of opportunistic goals attained by acquiring unique knowledge about a business, an industry or perhaps the world. While in some cases the output of a data science project is deployed in a production application, in many cases the result of a data science project is a paper accompanied by a document or PowerPoint slides to discuss insights that have business impact.

Planning vs. experimenting– There will no longer be a “project plan,” though there may be cycles that resemble agile framework sprints. Data science projects need to be managed like R&D projects following a scientific process that may introduce some perceived chaos and a high degree of uncertainty which is often difficult to manage. In reality, experiments are carefully designed and documented, with effort and timelines estimated. They are executed with a high degree of discipline and commitment.

Mitigating vs. embracing risk– Instead of identifying and mitigating risk factors, data science leaders should accept and manage uncertainty. Risk needs to be quantified and embraced before tackling a problem. Data science project leads should be very proactive in the beginning of a project in defining and quantifying the acceptable margin of error for a data science model to be successful.

Managing time and budget– Limiting time and budget may have a direct impact on insight quality. Insights can’t be scheduled on a calendar and the innovation process is neither linear nor progressive. The first experiment may yield the most accurate model but there is no way of knowing that without completing the rest of the experiments. However, most organizations have a limit of budget, time, or both. Time and budget will be maximized by clearly setting milestones and designing each experimental phase with the goal of achieving milestones. Data scientists are motivated to run experiments and find answers as quickly as possible. After exploring the data and running a few experiments, a good data science team will have a sense of the feasibility of reaching the milestones.

Providing ongoing maintenance – To realize the full potential of artificial intelligence, data science models need to be continuously fed new data, re-trained, so that they can get smarter and provide more accurate predictions and better insights.

What Has Not Changed

Though the data science project management approach is quite different from that of traditional software development, there are many basic elements that have not changed:

It starts with choosing the right problem to solve– This remains the hardest barrier to beat. Ask the questions and then see if AI provides the right solution. Organizations getting caught up in the hype create a storm when trying to figure out how to inject AI into the organization before asking the right business questions first.

Projects are driven by milestones– Whether through an experimental process or scrum sprints, all project tasks are structured around achieving a clear set of milestones.

Communication and executive support are critical for success– As a leader, communicating vision, expectations, learnings, barriers and opportunities will be critical to managing successful data science projects. Support from executive leadership is key to getting buy-in from the rest of the organization.

Implementing change requires managing change– A lot of people resist change, even with something as exciting and promising as AI. People may worry about losing their jobs; they may not trust insights provided by an abstract mathematical model only a handful of people understand; or they may need to be trained to use new digital products. Change management strategies must be put in place to adopt AI across business units in an organization.

Managing data science teams is an art and a science. As shared in a Harvard Business Review article, humans are at the heart of every technology project. Managing technical and analytical resources is very challenging, especially if you do not have the technical acumen to pose the right questions. Managing people, motivating your team, and communicating clearly and often are traditional skills that are not growing old.

Why the Data Science Process Is Misconstrued

Data science is driven by an experimental process and this implies that the exact results cannot be guaranteed. This is often misconstrued as budget being potentially thrown down the drain if the experiments don’t go as planned. The level of uncertainty and experimentation will vary depending on the problem that is being solved. Object detection models and basic chatbots, where the technology is more advanced and widely used, bear a minimal amount of risk compared to building a self-driving car which requires much more research. When tackling problems that require a higher level of experimentation and research, while the exact outcome is not guaranteed, valuable insights are always derived after each phase in the data science process. The outcome may be a 90% accurate model, or it may be insights on additional data that may need to be collected to be able to generate answers to questions through an AI model. If there is value in determining an organization’s capacity to extract insights and make progress toward achieving those milestones, then the investment in AI will be worth every penny.

The Data Science Process

The first widely used data science process (or back then, data mining process) was Cross-Industry Standard Process for Data Mining (CRISP-DM) which was introduced in 1996. It included six major phases:

  • Business Understanding
  • Data Understanding
  • Data Preparation
  • Modeling
  • Evaluation
  • Deployment

CRISP-DM evolved and a new data science framework called OSEMN (Obtain, Scrub, Explore, Model, Interpret) emerged in 2010. Most data science teams implement some variation of it.

Data science managers leading innovation projects that need to be deployed in a production system should create their own flow based on lessons from industry experts that align with their processes and governance structure. A suggested workflow to evaluate is Github’s machine learning team described in the O’Reilly publication: Development Workflows of Data Scientists. A lot of data science projects result in presenting insights in papers or slide decks, but the most advanced in the industry take those insights and apply them to a machine learning model and ultimately into a live software application or business process.

Our team at Wovenware extends the OSEMN process by incorporating an AI strategy design. The high-level steps are the following:

  1. Define Problem- What problem will be addressed? What is the business impact?
  2. Define Success Metrics– How will success and failure be tracked and measured?
  3. Gather Data – Identify data sources, define inputs and collect data.
  4. Cleanup & Process Data– Perform data cleansing, scaling, normalizing.
  5. Explore Data– Analyze the data and identify subgroups, outliers, tendencies.
  6. Identify Features – Form a hypothesis and identify features to be used.
  7. Prototype– Build exploratory models and revise the problem and features, iterating as needed.
  8. Build Infrastructure– Build and test the infrastructure for the model.
  9. Operationalize Model– Gather new data, develop integration pipelines, retrain and optimize models iterating as needed.

Operationalizing Data Science

The most challenging and exciting part of data science project management is taking an abstract mathematical model and integrating it into an existing software product for the world to use. Creating this link not only between the technologies, but between the teams and business units involved is a journey and a process. As the innovation process matures, productionizing a model will require implementing a more traditional and operational process that includes a timeline, budget, and project schedule. Employing traditional project management skills will be critical

The AI revolution is exciting. Executing data science projects with the right balance of innovation, experimentation, planning, research and discipline is key for managers shifting from traditional software development to data science project management.

Building Data Science Teams

Despite concerns of killer robots taking over the world, research shows that adoption of Artificial Intelligence (AI) across the industry is accelerating at a rapid pace. In 2019, McKinsey reported that AI adoption has increased in most industries, with 58 percent of organizations participating in the survey reporting that they have implemented at least one AI capability, up from 47 percent in 2018. Given this trend, traditional software organizations need to be rewired in order to build high-performing data science teams that can drive AI innovation.

Rewiring the Organization

Traditional software organizations are hard-wired to build software applications that are usually driven by efficiency. Most “smart” applications built in this century have translated business processes into standardized and deterministic rules that can be automated by code. By adopting AI, rules-based applications will now be able to process knowledge and data and generate valuable insights for a business.

Realizing AI’s full potential value goes above and beyond adopting a new technology. It requires shifting mindsets across departments in the entire organization, starting at the very top. Kathryn Hume did a great job of explaining it in a recent AI at Work Podcast: “The opportunity in running an AI-first organization is to shift the mindset around what a business process is — from a vehicle to drive standardization and efficiency, to a vehicle to collect unique knowledge about the world that can only be known via the processes that you have and the customers that you have.” Transforming efficiency- driven companies into knowledge-driven companies will undoubtedly change the way every department works. For many companies, one of the first steps in this transformation is building a data science team.

The First Recruits

Incorporating statistical analysis and applied mathematical models into business operations and products is an age-old practice. Financial services and insurance companies have been employing quantitative analysts for decades. Companies in other industries that are jumping on the AI bandwagon are following suit by hiring a lead data scientist to spearhead innovation efforts.

When building data science teams, the first recruit will likely have a couple of years of experience in the industry and a formal Master’s or Ph.D. degree in computer science, engineering, mathematics, physics, data science or machine learning. Although it is very common for organizations to hire experienced data scientists, Andrew Ng, founder of, suggests in a recent interview with Forbes, that hiring a couple of engineers to work on a small project is a good way to get the wheels turning and get a feel for what AI can do before defining a strategy for the organization.

However, not even the most talented data scientist will be able to get the job done alone. Different from the traditional quantitative analysts (quants) who have worked in banks and insurance companies, moving machine learning models into existing software applications requires close collaboration between people with a very diverse set of skills. The importance of people with industry, domain, and business knowledge is often underemphasized. Domain and business experts have a critical role in selecting the right questions to ask, identifying the untapped knowledge that would create the most impact, and interpreting the insights extracted from AI models.

The Economist defines the new craftmanship of data scientists as “the combination of the skills of software programmer, statistician and storyteller/artist to extract the nuggets of gold hidden under mountains of data.” To get the job done, data scientists will need a clear direction on the “nuggets of gold” and how to find answers to questions they are searching for.

Artificial Intelligence Team Roles

Though currently many companies struggle to hire data scientists, building an AI-first company or an AI product requires teamwork and a more diversified set of skills. The key roles in high performing AI teams which are summarized by this Forbes article include the following:

  • AI project manager: The project manager coordinates projects across all job functions and helps a company identify the business challenge it is hoping to solve through AI. They are the link between business leaders and the data science team.
  • Data engineer: The data engineer is responsible for gathering the data and supplementing it with external data if needed. They have specialized business knowledge as well as technical expertise with data models. They analyze and manipulate data so that it is in an optimal clean and normalized state and can be fed to AI models.
  • Data scientist: Data scientists are generally mostly focused in academic research, forming hypothesis, & testing statistical models and algorithms. Ph.D candidates focus on the mathematical theory rather than the applications in industry.
  • Machine learning engineer: Machine Learning Engineers- ML Engineers integrate mathematical models into live industry applications and products. They combine an expertise on AI models with experience on designing and programming applications and architectures that scale.
  • AI ethicist: The role of AI ethicist is still evolving. It emerged to meet the need of ensuring that AI models address any underlying bias in the data and enforce fairness throughout the AI lifecycle.

While there is general agreement that AI teams must have representation of the roles described above, there is no consensus of the right proportions for each role. In the beginning, it is likely that one person may fulfill multiple roles. As the organization grows and evolves in AI adoption, so will the roles within the team. Some organizations that start with a team of three data scientists, may end up with a team of two data scientists and three machine learning engineers. Likewise, other organizations that rely heavily on engineers may evolve to hire more data scientists. The right proportions can vary greatly, and each organization will adjust and evolve to find its own unique path.

Hiring vs Training Your Own

Here at Wovenware we’ve built an AI team from the ground up, hiring experienced workers and training new hires and experienced software developers. There is some hype in the industry about hiring only Ph.D level data scientists. They are often treated as unicorns. But it’s important to nurture a a keen sense for identifying talent with the right quantitative and analytical skills. Those who are hungry for solving complex problems will jump at the opportunity to up-skill. After going through both academic and on-the-job training Wovenware trainees learn how to read research papers, conduct experiments, build AI models and communicate insights with effective storytelling techniques. They are adequately equipped with the knowledge and tools to solve AI problems.

Creating both broad and specialized training programs is a wise move for any organization looking to build internal data science teams. Experience shows that resources having formal academic training do not always have an advantage over resources that learn using a combination of online programs and on- the- job training. For example, in an earlier post we’ve outlined some of the courses Wovenware software developers in training have taken to become machine learning engineers. A strong internal team will combine formally trained new hires with internal talent that has deep knowledge about your organization and applications.

Data Science Team Structure

Companies are implementing three main business models to integrate data science teams with the rest of the organization:

  • Centralized: Projects across the organization are handed off to a central data science team.
  • Decentralized: Business units across the organization are assigned dedicated data scientists and engineers.
  • Hybrid: Business units have designated AI experts who work with them on a daily basis but they also report to and rely on a centralized data science team to execute projects.

Each business model varies in the way and the frequency with which data science team members interact with business units and among themselves. It is fairly common to have reporting structures evolve over time for the same company.

Partnering with External Teams

Whether an AI Team is centralized or decentralized, every organization should have access to external teams that can extend and augment their internal team’s competencies or add velocity to their development. The staffing requirements and technology infrastructure required to develop scalable AI solutions are often beyond the reach of many organizations. As explained in an earlier post, by turning to AI outsourcing and nearshoring, organizations can get the high-quality AI solutions they need cost-effectively.

Getting Started

Andrew Ng gives companies insightful advice on where to begin when it comes to building an AI team. He says, “Figure out how to jump in. Even if it’s just a junior programmer on small projects, get the wheels going and feel what an AI application can do. Companies tend to want strategies around everything. But if a company has never done an AI application, they can’t strategize properly, so C-suites develop strategies that look cut and pasted from a newspaper headline. Someone else’s strategy is rarely right for them. So, I say: Just hire a couple of engineers to see what they can do, and keep growing from there.”

Wovenware COO, Carlos Meléndez also suggests nearshoring as a fast path and short cut to AI deployment. By deploying AI as a service – paying as you go for the development of specific algorithms, companies can forego the costs of building an AI infrastructure hiring dedicated data scientists and developing private crowds. In an age when “build-your-own,” is being replaced with “as a service,” nearshoring is gaining renewed traction as a speedy vehicle to get on the AI fast track.

Scaling the Team

Building data science teams requires a combination of aligning the corporate culture and mindset with that of a knowledge-driven economy, providing executive sponsorship for innovation initiatives, finding the right talent, and combining cross-functional and technical expertise in the team. The success of artificial intelligence lies in looking beyond the data science role and following a holistic and practical approach to drive greater innovation.