As fate willed, I was lucky for the last 1.5 - 2 months to analyze the Data Science and Machine Learning market. And there was a desire to write at least a few lines about this. So it will most likely be a short note rather than a solid article.
Marketing departments in large companies
Data Science is in demand here for the analysis of sales, markets, customer segmentation and research of their behavior. Most likely, in this case, the analyst's work is largely reduced to data aggregation, building dashboards and setting up marketing campaigns. The last task is an optimization task, within which models are built that predict the client's reaction to advertising, all kinds of promotions and discounts during the sale.
Risk management in banks and insurance companies
Risk management is actually a separate industry, and quantitative analytics is a tool for controlling and managing risk. The most common topic here is credit risk management, assessing the creditworthiness of borrowers and advanced methods for calculating regulatory and economic capital and reserves for expected losses from loan defaults.
R&D - laboratories
R&D - Research and Development. As part of the work of such laboratories, fundamental research is often carried out with the development of new algorithms and machine learning architectures. The specialists that are required in such areas are much more specialized and “digging” deeper than the classic data scientists. R&D professionals often refer to themselves as machine learning or deep learning engineers, or simply mathematicians.
Here are some examples of tasks: development of trainable agents in computer games, control of robotic equipment, unmanned aerial vehicles, autonomous driving.
Product startups
The era of startups is not over yet. The topic of launching startups and venture investing is still popular. The main feature of this area is the focus on a whole product. Machine learning, if used, the main factor of its usefulness is increasing the usability of the product and user experience (UX).
For example, a mobile augmented reality application for children. The popularity of such an application may largely depend not on the “soulless” quality metric, but on the brightness and spectacularity of the picture. Another example: a chatbot for teaching English or just for fun. Quality metrics are far from obvious. The chatbot can speak out of the topic, but it will sound "cool" and gain views, clicks and likes. It is not hard to guess where I am leading here. Such applications or sites can make money at least on advertising.
Integrators, IT consulting
Integrator companies and consulting services are in demand as they aggregate experience and knowledge. Their main value lies in human capital. To launch any project with automation and machine learning requires the knowledge of many professionals from completely different fields at once. No one person is able to combine both expertise in the best industry practices (banking, retail, advertising, social media) and knowledge of the entire technology stack. A striking example is the MLOps practice (a subset of DevOps practices as applied to machine learning) offered by Neosoft. An alternative way to take your business to the next level is to hire an entire team, and this is done in just two clicks of fingers.
Vendors and software developers
Business automation and modernization is built on the one hand on ready-made solutions, and on the other hand, it cannot do without customization. The task of customization for a specific infrastructure and business model, of course, can also be solved by integrators at the level of settings for the purchased ready-made software. But often, in order to gain a competitive advantage, a company must convey to the market some of its unique service or product offer. For example, third-party developers like EPAM are attracted even by companies like Google or Facebook.
Tech IT companies giants and platforms
Among the tech giants, of course, you can name "search engines" (Google, Yandex), online commerce (Amazon, Alibaba), social networks (Facebook, Instagram, WeChat). These guys, if they need something, often buy startups and companies entirely and make their own internal structural divisions out of them.
A steady trend in recent years is associated with the transition of everything and everyone to cloud platforms. In this connection, entire ecosystems of partner services are being built based on platforms such as Azure, AWS or Google Cloud. In particular, these services offer customized access to machine learning and data mining capabilities.
Outcome
In order to survive in the existing variety of offerings on the market, any company must clearly understand what areas of business automation, machine learning and data analytics it specializes in. It is also very important to know the best industry practices and trends of your client, and of course your competitors. But the most important thing is that the client knows and recognizes you.