For many aspiring data scientists, the dream job is in one of the global tech companies.

对于许多有抱负的数据科学家而言，理想的工作是在全球高科技公司之一中。

But when you focus on the tech industry only, you exclude an enormous number of cool jobs in other sectors.

但是，当您仅关注技术行业时，您会排除其他行业中的大量出色工作。

I have worked for many different industries, and I would not like to miss the experience I have made. In my former position, I worked on projects in the chemical industry. I did not know that industry before, and I was surprised how advanced they are, and the many insights I could gain, especially the integration with IoT technology.

我曾在许多不同的行业工作，我不想错过自己的经验。 在我以前的职位上，我从事化工行业的项目。 我以前并不了解该行业，我感到惊讶的是它们的先进程度以及我可以获得的很多见解，尤其是与IoT技术的集成。

So, what are the reasons to look for a job in another industry?

那么，在另一个行业找工作的原因是什么？

1. You work on exciting projects that interconnect with other technologies like genetics and IoT
   您正在从事与遗传学和物联网等其他技术互连的令人兴奋的项目
2. You work with a massive amount of data from different sources like machines, manufacturing processes, and wearables
   您需要处理来自不同来源(如机器，制造过程和可穿戴设备)的大量数据
3. You are living in a location where no tech company is nearby
   您住在附近没有科技公司的地方
4. You just do not like to work in the tech industry
   您只是不喜欢在科技行业工作
5. You already have another industry background, and you would like to apply your knowledge. Further, an internal move is incredibly attractive as it increases the chance to get hired because of the specialized industry knowledge
   您已经具有其他行业背景，并且想应用您的知识。 此外，由于专业的行业知识，内部举动极具吸引力，因为它增加了被录用的机会

In the following, I give you five sectors to find exciting job positions outside of the tech industry. I will provide you with a brief overview of the industry, the salary level compared to the tech industry, how recession-proof this sector is, examples of projects, and what skills the industry is looking for in data scientists.

在下文中，我将为您提供五个部门，以找到技术行业以外的令人兴奋的工作职位。 我将向您简要概述该行业，与高科技行业相比的薪水水平，该行业如何抗衰退，项目示例以及该行业在数据科学家中寻求的技能。

生命科学与制药 (Life Sciences and Pharmaceuticals)

Industry

行业

COVID-19 forces us to hear all the time about the development of vaccines. That shapes our view of this industry. But this is only one part of the whole sector.

COVID-19迫使我们一直听到有关疫苗开发的信息。 这塑造了我们对该行业的看法。 但这只是整个行业的一部分。

The fact is that the industry is big and very heterogeneous, with many subsectors like agricultural bioscience, diagnostics, therapeutics, pharmaceuticals, genomics and proteomics, veterinary life sciences, cosmetics, drug manufacturing, medical technology, distribution, and many other services.

事实是，行业大而非常不均匀，有许多子行业像农业生物科学，诊断，治疗，药品，基因组学和蛋白组学，兽医生命科学，化妆品，药品制造，医疗技术，销售等多方面的服务。

In the pharmaceutical sector alone, we have, besides vaccine development, other areas like antibodies therapeutics, cell therapies, generic medicines, immunotherapy, proteins, or stem cells.

仅在制药领域，除了疫苗开发以外，我们还拥有抗体治疗，细胞疗法，非专利药物，免疫疗法，蛋白质或干细胞等其他领域。

The costs for developing new treatments start today from around one billion U.S. dollars and go up to two-digits billion U.S. dollars. The industry makes all efforts to decrease the costs and make treatments faster available. That pushes for a data-driven sector comparable to the technology industry. All global players are building data ecosystems to link the patients, the service providers like hospitals and doctors, and their own data together.

开发新疗法的费用从今天的约10亿美元开始，上升到两位数的十亿美元。 业界竭尽全力降低成本并加快治疗速度。 这就推动了一个与技术行业可比的数据驱动行业。 所有全球参与者都在建立数据生态系统，以将患者，医院和医生等服务提供商以及他们自己的数据联系在一起。

I advise you to perform some research about the different subsectors. When you see a job offer, always find out the subsector first.

我劝你还是执行有关不同的一些研究分部门 。 当您看到工作机会时，请务必先找到该子行业。

The industry typically has a low staff turnover and offers excellent benefits, with high salaries in the same range as the tech companies.

该行业的员工流失率通常较低，并且会提供出色的收益，其高薪与科技公司的工资相同。

Recession-proof

防衰退

Yes, the industry is very recession-proof. Health care is always needed, for humans as well as for pets and animals. And in times like today, the industry gets even a boost.

是的，该行业非常抗衰退。 对于人类以及宠物和动物，始终需要医疗保健。 在今天这样的时代，该行业甚至得到了提振。

Projects

专案

You can imagine that the range of possible areas and projects is as wide as the number of subsectors. A comprehensive list of projects would fill pages. So, I limit it on three examples.

您可以想象，可能的领域和项目的范围与子部门的数量一样大。 一份完整的项目清单将填满页面。 因此，我将其限制为三个示例。

Precision medicine: Medical treatments are more and more based on individual, personalized characteristics of a patient. These characteristics are disease subtypes, personal patient risks, health prognosis, and molecular and behavioral biomarkers. A biomarker is any measurable data point such that the patients can be stratified, e.g., disease severity score, lifestyle characteristics, or genomic properties. Based on all these data, the best treatment of a single patient is determined.

精密医学：越来越多地根据患者的个性化特征来进行医疗。 这些特征包括疾病亚型，个人患者风险，健康预后以及分子和行为生物标志物。 生物标志物是任何可测量的数据点，以便可以对患者进行分层，例如疾病严重程度评分，生活方式特征或基因组特征。 基于所有这些数据，确定单个患者的最佳治疗。

A real example is a patient with ovarian cancer where chemotherapy was not effective. So, one performed a genome sequencing for finding the misplaced nucleotide bases that cause that cancer. With big data analytics, one found amongst the 3 billion base pairs of a human the modification (this corresponds to the number of words of 7798 books of Harry Potter’s The Philosopher’s Stone). This modification was known from lung cancer, where a drug exists. This drug was applied, and the patient recovered.

一个真正的例子是化疗无效的卵巢癌患者。 因此，人们进行了基因组测序，以寻找引起癌症的错位核苷酸碱基。 通过大数据分析，在人类的30亿个碱基对中发现了一个修改(这对应于哈利·波特的《哲学之石》的7798本书的单词数)。 从存在药物的肺癌中知道这种修饰。 应用了这种药物，病人康复了。

Supply chain optimization: The production of drugs needs time, especially today’s high-tech cures, based on particular substances and production methods. Also, a drug can be stored only for a limited time, and some need special storage, e.g., in a cold storage room. The whole planning from having the right input substances available at the right time, having the adequate production capacity, and finally, the proper amount of drugs stored for serving the demand, is a very complex system. And there is not only one drug where this must be managed. No, there are hundreds and thousands of them, each with their specific conditions. So, life sciences companies are starting to manage their whole supply chain with data science methods.

供应链优化：药物的生产需要时间，特别是基于特定物质和生产方法的当今高科技疗法。 而且，药物只能存储有限的时间，有些药物需要特殊存储，例如在冷藏室中。 从正确的输入物质在正确的时间可用，具有足够的生产能力以及最终存储适当数量的药物来满足需求的整个计划是一个非常复杂的系统。 而且不仅有一种药物必须加以管理。 不，它们有成千上万，每个都有其特定的条件。 因此，生命科学公司开始使用数据科学方法来管理整个供应链。

Research publication scanning for biomarkers: Daily, there are many hundreds of scientific publications about the detection of biomarkers for all the different diseases by the multiple research teams around the globe. It is vital for life sciences companies to know the newly detected biomarkers to implement them into their particular research fields to find new cures. Research is enormously expensive and time-critical. By understanding the current relevant research, they can avoid duplication in research, and they can speed up the time to market. The amount of information is so enormous that this cannot be done manually. So, very sophisticated NLP algorithms are developed that find the relevant publications. Besides the understanding of the content where such biomarkers are relevant, a judgment of the quality of the published results and how it fits into the company’s research must be given — a very complex task.

扫描生物标志物的研究出版物：每天，全球有数百个研究团队针对数百种关于所有不同疾病的生物标志物检测的科学出版物。 对于生命科学公司而言，了解新发现的生物标记物以将其应用于特定研究领域以寻找新的治疗方法至关重要。 研究是非常昂贵且时间紧迫的。 通过了解当前的相关研究，他们可以避免重复研究，并可以加快产品上市时间。 信息量巨大，以至于无法手动完成。 因此，开发了非常复杂的NLP算法来查找相关出版物。 除了要了解与这些生物标志物相关的内容之外，还必须对已发表结果的质量以及其如何适合公司研究进行判断-这是一项非常复杂的任务。

Skills needed

所需技能

The job typically requires a science background in addition to data science knowledge. For senior positions, the proven experience in working with health records data is needed. For specialized topics, the companies can be very picky with the matching skills. All expertise from working in a complex scientific or engineering environment is highly welcomed. Knowledge in bioinformatics is a plus. Especially for people with a science background that want to enter the data science field, this industry gives plenty of opportunities.

除数据科学知识外，该工作通常还需要科学背景。 对于高级职位，需要在处理健康记录数据方面具有可靠的经验。 对于专业主题，公司可能对匹配技巧非常挑剔。 我们欢迎在复杂的科学或工程环境中工作的所有专业知识。 具有生物信息学知识者优先。 特别是对于那些具有科学背景的人想要进入数据科学领域，这个行业提供了很多机会。

实用工具 (Utilities)

Industry

行业

The utilities industry is providing basic public amenities like water, electricity, or natural gas. They build, own, and maintain the corresponding infrastructure like hydroelectric generators, nuclear plants, energy grids, control stations, water distribution systems, etc. That defines the supply. On the other side of the equation is the demand of consumers, i.e., single persons and corporates, which has to be served.

公用事业行业正在提供基本的公共设施，例如水，电或天然气。 他们建立，拥有和维护相应的基础设施，例如水力发电机，核电站，能源网格，控制站，配水系统等。 另一方面，必须满足消费者的需求，即单身人士和公司的需求。

The interaction of supply and demand and infrastructure maintenance is quite complex, and today heavily technological and real-time data-driven. The industry is moving fast into IoT devices and so-called smart metering, which records real-time any consumption of energy or water and provides detailed information to the utilities and the consumer. It is considered as the first step into smart grids.

供需与基础设施维护之间的相互作用非常复杂，如今，技术和实时数据受到大量驱动。 业界正快速进入IoT设备和所谓的智能计量，该技术可实时记录能源或水的任何消耗并向公用事业和消费者提供详细信息。 这被认为是进入智能电网的第一步。

This industry operates in two extremes: on the one hand, they have to maintain assets with a lifetime from 35 years, like generating plants up to 100 years for drinking water distribution systems. On the other side, electricity cannot be stored, and real-time management is required.

这个行业有两个极端情况：一方面，他们必须维护使用寿命长达35年的资产，例如为饮用水分配系统生产长达100年的电厂。 另一方面，电力无法存储，因此需要实时管理。

Utilities are one of the industries with the highest volume of data, and thus, they face all related challenges to get usage of them. The industry is currently also in a technological transformation and adds more and more sensors and IoT devices, resulting in even more available real-time data.

公用事业是数据量最大的行业之一，因此，要使用它们，它们会面临所有相关挑战。 该行业目前也在进行技术改造，并添加了越来越多的传感器和IoT设备，从而产生了更多可用的实时数据。

Based on my experience, the salary is about 5–10% lower on average than in the tech industry.

根据我的经验，薪水平均比科技行业低约5-10％。

Many utilities have outsourced their data science capabilities. So, check for specialized small and middle-sized consulting firms that serve purely the utilities market.

许多公用事业公司已将其数据科学功能外包。 因此，请检查专门为公用事业市场服务的专业中小型咨询公司。

Recession-proof

防衰退

Yes, the industry is very recession-resistant. Water, electricity, and power are always needed, independent of the economic state.

是的，该行业非常抗衰退。 始终需要水，电和电，而与经济状况无关。

Projects

专案

There are many areas in the utilities space where data science is used, from asset performance management, outage, and failure prediction, to the energy supply management and customer analytics.

公用事业领域中有许多使用数据科学的领域，从资产性能管理，中断和故障预测到能源供应管理和客户分析。

Water leakage detection: In Europe and the U.S., trillions of liters of drinking water leak out of the water supply system. The installation of water sensors to detect leakage is still at the beginning. Manual measurement and tapping of pipes are standard but neither efficient nor region-wide applicable. So, based on the flow measurements at specific locations and the end consumers’ water consumption data, prediction models are set up to localize areas of potential water loss.

漏水检测 ：在欧洲和美国，数万亿升的饮用水从供水系统中漏出。 仍然开始安装用于检测泄漏的水传感器。 手动测量和分接管道是标准配置，但既无效率，又不适用于整个地区。 因此，根据特定位置的流量测量结果和最终用户的用水量数据，可以建立预测模型以定位潜在的水损失区域。

Preventive power grid management with drones: Power grids have alone in the U.S. a length of 160,000 miles for high-voltage power lines and millions of miles for the low-voltage local distribution lines. Monitoring all lines for potential risks like, e.g., trees that could damage a line, is a nearly impossible undertaking. So, commercial drones are used to patrol along the lines and record videos of the adjacent environment. Computer vision algorithms are developed or improved for the automated detection of potential risks.

用无人机进行预防性电网管理：仅在美国，电网就拥有160,000英里的高压电力线长度和数百万英里的低压本地配电线长度。 监视所有生产线是否存在潜在风险，例如可能损坏生产线的树木，这几乎是不可能的。 因此，商用无人机用于沿线巡逻并记录邻近环境的视频。 开发或改进了计算机视觉算法，用于自动检测潜在风险。

Consumption forecasts and dynamic pricing: The prediction of electricity consumption is one of the most important figures to ensure the supply. Remember, energy cannot be stored. So, in real-time, the supply must be guaranteed. That is a very complex system, from managing hydroelectric generators’ production capacity to the distribution through the power lines, purchasing and selling electricity, and setting incentives to use less or more power during specific time frames with the corresponding pricing. You must manage overlapping long-term and short-term effects, several seasonalities, weather forecasts, and short-term fluctuations. It is a very delicate issue and not easy to solve with all the data volume and a short time to react to the predicted outcome.

消耗量预测和动态定价：用电量的预测是确保电力供应的最重要数据之一。 请记住，能量无法储存。 因此，必须实时保证供应。 这是一个非常复杂的系统，从管理水力发电机的生产能力到通过电力线进行配电，买卖电力，并制定激励措施，在特定的时间范围内以相应的价格使用更少或更多的电力。 您必须管理重叠的长期和短期影响，几个季节，天气预报和短期波动。 这是一个非常棘手的问题，要解决所有数据量和对预测结果做出React的时间很短，这不容易解决。

Skills needed

所需技能

The utilities industry is working with a lot of mathematical models, mainly from Operations Research. The data scientist should therefore have a good foundation in mathematics and mathematical models. You should have at least basic knowledge of how the industry works and how to proceed a large amount of data, even in real-time.

公用事业行业正在使用许多数学模型，主要是来自运筹学。 因此，数据科学家应该在数学和数学模型方面有良好的基础。 您至少应具有该行业的运作方式以及如何处理大量数据的基本知识，甚至是实时的。

餐饮业 (Food and Beverage Industry)

Industry

行业

The food and beverage industry contains a wide range of companies from restaurants, coffee shop and fast-food chains, food, and beverage transportation services, food manufacturer, to the giant multinationals like Nestlé, PepsiCo, Anheuser-Busch InBev, JBS, and so on.

食品和饮料行业包含众多公司，从餐馆，咖啡店和快餐连锁店，食品和饮料运输服务，食品制造商到雀巢，百事可乐，百威英博，JBS等大型跨国公司上。

My recommendation deals primarily with the many opportunities in the multinationals. They have several brands, functional food and drinks, and sometimes pet food. Their business does not only depend on consumer behavior but also all the actions of sellers of their products.

我的建议主要涉及跨国公司的许多机会。 他们有几个品牌，功能性食品和饮料，有时还有宠物食品。 他们的业务不仅取决于消费者的行为，而且还取决于产品卖方的所有行为。

The salary level of the multinationals is comparable to the tech industry. Local companies pay considerably lower salaries.

跨国公司的薪水水平与科技行业相当。 当地公司支付的工资要低得多。

Recession-proof

防衰退

The industry is partially recession-proof. Both, soft drinks and alcoholic beverages like wines and spirituous beverages, see declines during recessions. In contrast, large brewers face only a minor fall. Restaurant spendings drop heavily during recessions, whereas grocery store spendings remain stable, and discount retailers can increase their sales.

该行业部分抵御衰退。 软饮料和含酒精饮料(如葡萄酒和烈性饮料)在衰退期间都会下降。 相比之下，大型啤酒厂仅面临较小的跌幅。 在经济衰退期间，餐厅支出大幅下降，而杂货店支出保持稳定，折扣零售商可以提高销售额。

Projects

专案

Sentiment analysis and critical commentary detection: Today, complaints about a product are first public on social media platforms. Severe cases like, e.g., contamination of food products, are swiftly in the headlines. For large corporates, it essential to monitor real-time 24/7 all their product comments around the globe. When an impactful comment is identified, the brand needs to act quickly. The background of the potential incident needs to be analyzed and corresponding actions initiated.

情绪分析和批评评论检测：如今，有关产品的投诉已在社交媒体平台上首次公开。 头条新闻Swift报道了诸如食品污染等严重案件。 对于大型公司，至关重要的是要实时监控全球24/7的所有产品评论。 识别出有影响力的评论后，品牌需要Swift采取行动。 需要分析潜在事件的背景并启动相应的措施。

Many negative sentiments need no intervention. On the other side of the scale are incidents that require immediate executive committee attention. Finding the right balance in the machine learning algorithms is tricky. It is a highly exciting field where misclassification matters.

许多负面情绪不需要干预。 另一方面，需要立即引起执行委员会注意的事件。 在机器学习算法中找到合适的平衡是很棘手的。 这是一个非常令人兴奋的领域，错误分类很重要。

On-time production and delivery: Soft drinks beverage companies need to plan the demand for drinks, calculate back their production schedule and capacities, and their ingredients demand from their suppliers. Besides seasonalities, factors like the weather, bigger and smaller events, consumer trends, price wars, economic situations, seller purchase power, production time and capacities, shelf life, etc., must be considered. The logistics need to be arranged in advance to ensure the delivery on-time of the drinks to sellers or event catering. It is again a non-trivial task and demands for highly-skilled data science teams.

准时生产和交付：软饮料企业需要计划饮料需求，重新计算生产时间表和产能，以及供应商对食材的需求。 除季节性外，还必须考虑天气，更大或更小的事件，消费者趋势，价格战，经济形势，卖方购买力，生产时间和产能，保质期等因素。 需要提前安排物流，以确保将饮料按时交付给卖家或活动餐饮。 对于高技能的数据科学团队来说，这又是一项艰巨的任务和要求。

Quality assurance: The consumer expects that the taste, consistency, quality, and shelf life of a product are consistent all over the year and independent of the consumption location. Many factors influence the production outcome, starting from the proper amount of ingredients, their quality, the season, the production country and transportation conditions, storage, and production conditions like temperature, pressure, variances in the production time, and so on. To find the right pattern of ingredients and production parameters per produced batch such that the product is always the same is a highly non-trivial data science task.

质量保证：消费者希望产品的味道，稠度，质量和保质期在一年中保持一致，并且不受消费地点的影响。 影响生产结果的因素很多，从适当的配料量，质量，季节，生产国家和运输条件，存储以及生产条件(例如温度，压力，生产时间的变化等)开始。 要找到正确的配料模式和每个生产批次的生产参数，以使产品始终相同，这是一项非常重要的数据科学任务。

Skills needed

所需技能

They expect in-depth data science method knowledge from regression to neural networks. When working on production problems, time series knowledge is needed, as well as NLP experience in the customer analytics field.

他们期望从回归到神经网络有深入的数据科学方法知识。 处理生产问题时，需要时间序列知识以及客户分析领域的NLP经验。

工业品 (Industrial Goods)

Industry

行业

In general, one can say that this industry contains all the production, manufacturing, equipment, and sales of them that are not sold directly to a consumer. Examples of the subsectors are construction, manufactured houses, industrial machines and tools, cement and metal fabrication, and industrial components. Examples of well-known global companies are GE, Siemens, Hitachi, 3M, Honeywell International, Bosch, Lockheed Martin, or ABB. Over the last years, these companies not only went through their own full digital transformation and have fully automated manufacturing, but they have their own platforms as a service for their customers and extensive data and technology teams.

通常，可以说这个行业包含了所有不直接出售给消费者的生产，制造，设备和销售。 该子行业的示例是建筑，房屋，工业机械和工具，水泥和金属制造以及工业组件。 全球知名公司的示例包括GE，西门子，日立，3M，霍尼韦尔国际，博世，洛克希德·马丁或ABB。 在过去的几年中，这些公司不仅进行了自己的全数字化转型并实现了完全自动化的制造，而且还拥有自己的平台，可为客户和庞大的数据和技术团队提供服务。

The industry has a particularity, the so-called “hidden champions.” A hidden champion is a company that is at the top in the global market but has revenues below $5 billion, and it is nearly unknown in public. These companies are typically high-tech firms, thought leaders in digitalization and technology, and usually located in nowhere. Data and AI is their driver. It is a paradise for a data scientist. However, to find such a hidden champion and get hired is not an easy task.

该行业具有特殊性，即所谓的“隐形冠军”。 一个隐藏的冠军是一家在全球市场上名列前茅但收入低于50亿美元的公司，在公众场合几乎是未知的。 这些公司通常是高科技公司，是数字化和技术领域的思想领导者，通常不在任何地方。 数据和人工智能是他们的驱动力。 这是数据科学家的天堂。 但是，要找到这样一个隐藏的冠军并被录用并不是一件容易的事。

The salary level is close to the level of the tech industry. Especially at hidden champions are the benefits and working culture above average, and the employee is still a valuable human and not a number.

薪水水平接近科技行业水平。 尤其是在隐性的领导者身上，福利和工作文化要高于平均水平，而员工仍然是有价值的人才，而不是数字。

Recession-proof

防衰退

The simple answer is: it depends. There is no general statement possible, and it depends on the subsector and the type of recession. To give you an example: in the current COVID-19 crisis, fashion production collapsed, and with it, the demand for industrial machines in that industry. But one hidden champion that is producing sewing machines cannot meet the global demand. The companies’ high precision machines are flexible in the application and are now used for masks production. My observation is that hidden champions are less likely to be hit by a recession.

简单的答案是：这取决于。 没有一般性的陈述，这取决于子行业和经济衰退的类型。 举个例子：在当前的COVID-19危机中，时装生产崩溃，随之而来的是该行业对工业机器的需求。 但是，一个隐藏的生产缝纫机的冠军无法满足全球需求。 该公司的高精度机器应用灵活，现已用于口罩生产。 我的观察是，隐藏的冠军不太可能受到经济衰退的打击。

Projects

专案

The industry and the projects are driven by the sensors, connected devices, and automation that make up intelligence machines, which produces an enormous amount of real-time data.

行业和项目由构成智能机器的传感器，连接的设备和自动化驱动，这些机器产生大量的实时数据。

Workers’ safety: Worker’s accidents are costly, not only of the accident itself but also because of the absence of the worker and reputational damage. Today, the workers in the heavy equipment environment are more and more equipped with sensors, vibrators, or integrated cameras. So, the data of all these devices are analyzed. One example is the analytics of real-time data on patterns that detect a risky situation for the employee and leads to an alert, e.g., a high-pitched sound. Another example is the pattern detection for unhealthy work behavior like ergonomic wrong lifting and subsequent instructions for the worker to improve.

工人的安全：工人的事故代价高昂，不仅是事故本身，而且是因为没有工人和声誉受损。 如今，重型设备环境中的工人越来越配备传感器，振动器或集成摄像头。 因此，分析了所有这些设备的数据。 一个示例是对模式的实时数据进行分析，这些模式可检测员工的危险情况并发出警报，例如声音高亢。 另一个示例是针对不健康工作行为的模式检测，例如人体工学错误举起，以及随后的指示以供工人改进。

Productivity optimization: We all have heard of predictive maintenance. With machine learning methods, the time points of machine failure are predicted, and the anticipated maintenance. The leading companies today can detect small anomalies already a few weeks in advance before a failure happens. That gives not only time for a scheduled maintenance but also to optimize the whole production over that period, including how eventually, by adjusting production parameters can avoid the predicted failure. These are high-dimensional problems to solve that require methods on the frontier of machine learning.

生产力优化：我们都听说过预测性维护。 使用机器学习方法，可以预测机器故障的时间点以及预期的维护。 如今，领先的公司可以在发生故障之前提前几周发现微小异常。 这不仅为计划的维护提供了时间，而且还可以优化该期间的整个生产，包括最终如何通过调整生产参数来避免预期的故障。 这些都是高维度的问题，需要机器学习前沿的方法来解决。

Research and Development (R&D): The intelligent machines need intelligent and robust algorithms to perform their tasks. An example is diagnostics machines where a camera is installed to monitor the control panel, i.e., how it is used by the workers and their hand movements. These data are coupled with machine performance and failure data. Based on the patterns, self-adjusting processes are integrated into the machine to ensure consistent work performance.

研究与开发(R＆D)：智能机器需要智能且强大的算法来执行任务。 诊断机器就是一个例子，其中安装了摄像头以监视控制面板(即，工人如何使用其及其手部动作)的诊断机。 这些数据与机器性能和故障数据结合在一起。 根据这些模式，自我调整过程被集成到机器中，以确保一致的工作性能。

Skills needed

所需技能

They are looking for well-skilled data scientists with knowledge in advanced methods as well as in engineering. The experience in handling high volume and real-time data is critical. The personality for working in a team and a smaller company is equally important.

他们正在寻找具有先进方法和工程知识的技术精湛的数据科学家。 处理大量实时数据的经验至关重要。 在团队和较小的公司中工作的个性同样重要。

农业产业 (Agricultural Industry)

Industry

行业

The industry covers the science, art, and business of livestock, aquaculture, i.e., fisheries, crops and plants, and forestry. We are typically most familiar with farming. Farming transforms into a technology-driven business, too. With fewer resources, agriculture has to produce more and more products and food. The technological development covers autonomous tractors and machinery, digitalized disease detection and pest management, better weather predictions, automated irrigation, and harvesting systems, and livestock health, to mention a few. That only can happen with large amounts of data and many smart algorithms.

该行业涵盖了畜牧业，水产养殖业(即渔业，农作物，植物和林业)的科学，艺术和商业。 我们通常最熟悉农业。 农业也转变为技术驱动的业务。 由于资源减少，农业不得不生产越来越多的产品和粮食。 技术发展包括自动拖拉机和机械，数字化疾病检测和病虫害管理，更好的天气预报，自动灌溉和收割系统以及牲畜健康。 只有大量数据和许多智能算法才能实现。

Quick analytics on Glassdoor gives an average salary range of about 10% less than a similar position in a tech company to a nearly equal level. But one has to consider the location differences, as many agricultural data science jobs are not in the big centers where tech companies are located.

在Glassdoor上进行快速分析可以使平均薪资水平比技术公司的类似职位低10％左右，几乎达到同等水平。 但必须考虑位置差异，因为许多农业数据科学工作不在科技公司所在的大中心。

Recession-proof

防衰退

This industry is a mixed blessing. Agriculture is mostly a low margin business, and any cut in the spendings for food during a recession has a direct cut impact on the farmers’ income. On the other hand, the same leads to more digitalization to become more efficient and effective. So, on the innovation side there, we typically do not see a lot of change, and it can be considered as quite recession-resistant.

这个行业是喜忧参半。 农业主要是低利润业务，在经济衰退期间任何粮食支出的削减都会直接影响农民的收入。 另一方面，这导致更多的数字化变得更加有效。 因此，在创新方面，我们通常不会看到很多变化，并且可以认为它具有相当的抗衰退能力。

Projects

专案

Precision farming is the keyword. Based on real-time data about crops, plants, soil, hyper-local weather conditions, temperature, moisture, collected by sensors in the fields, satellite and drone images, individual plants’ needs can be analyzed and the corresponding care determined.

精确农业是关键词。 根据田间传感器收集的有关作物，植物，土壤，超局部天气状况，温度，湿度的实时数据，卫星和无人机图像，可以分析单个植物的需求并确定相应的护理措施。

The health of cows: Earmarks that measure temperatures, fitness trackers for cows, GPS-neckbands, and even sensors in the stomach are today’s available accessory for cows. These real-time data are then analyzed by machine learning algorithms to detect anomalies in the individual cow’s health, and a corresponding alert is then sent either to the farmer or directly to the veterinary. Such algorithms’ development is tricky because too many alerts or the wrong alerts would lead to incorrect treatments, and no warning in a severe health case would lead to no treatment.

母牛的健康：今天，可用于测量母牛的配件包括用于测量温度的标记，用于母牛的健身追踪器，GPS颈带，甚至是胃中的传感器。 然后，通过机器学习算法对这些实时数据进行分析，以检测个体母牛的健康状况是否异常，然后将相应的警报发送给农场主或直接发送给兽医。 此类算法的开发非常棘手，因为过多的警报或错误的警报将导致错误的治疗，在严重的健康情况下没有警告将导致无法进行治疗。

Disease detection and management: Either with images or infrared pictures of the fields, microscopic diseases can be detected with computer vision. Combined with further local data like weather, or soil parameters, the spread of a disease can be predicted. Based on all these data, the right treatment for precisely the contaminated area is determined. And it should ensure that neither too much nor too few pesticides are used. Here, you are working on the frontier of image recognition techniques.

疾病检测和管理：可以使用现场的图像或红外图像，通过计算机视觉检测微观疾病。 结合天气或土壤参数等其他本地数据，可以预测疾病的传播。 根据所有这些数据，确定正确的污染区域的正确处理方法。 并且应确保既不使用过多农药，也不使用过少农药。 在这里，您正在研究图像识别技术的前沿。

Harvesting and grading: Today, automated harvesting machines exist that also automatically grade fruits or vegetables. One example is tomatoes. They are often cultivated in greenhouses with robots. These robots not only manage data-driven cultivation but also harvest the ripe fruits and grade them into the different classes. Again, this is very machine learning intense work where you can use all the advanced neural network models.

收割和分级：今天，存在自动收割机，也可以对水果或蔬菜进行自动分级。 一个例子是西红柿。 它们通常是用机器人在温室中栽培的。 这些机器人不仅可以管理数据驱动的栽培，还可以收获成熟的果实并将它们分级。 同样，这是非常机器学习的工作，您可以在其中使用所有高级神经网络模型。

Skills needed

所需技能

The good message is that the agricultural industry does not expect that you have a corresponding education. They are looking for data scientists open and curious to expand the applications of data science and AI into less considered sectors. But do not underestimate the maturity of methods used. Be prepared for very advanced applications.

好消息是，农业行业不希望您接受相应的教育。 他们正在寻找开放和好奇的数据科学家，以将数据科学和AI的应用扩展到考虑较少的领域。 但不要低估所用方法的成熟度。 为非常高级的应用做好准备。

连接点 (Connecting the Dots)

Many exciting sectors outside of the tech industry give unique opportunities to acquire skills and expertise that you otherwise could not experience. Technology is penetrating all industries and sectors, and with that comes the generation of huge volumes of data. All sectors have to become data-driven.

科技行业以外的许多激动人心的领域为获得您以前无法体验的技能和专业知识提供了独特的机会。 技术正在渗透所有行业和部门，随之而来的是生成大量数据。 所有部门都必须成为数据驱动的。

My advice for you is that when you are looking for a data science job, be open-minded and think outside of the box. It will give you a competitive advantage for your data science career.

我对您的建议是，当您在寻找数据科学工作时，请放开胸怀，跳出框框思考。 这将为您的数据科学事业提供竞争优势。