Storytime

My Internship Journey

In 2022, I embarked on my journey as a Data Scientist intern at a vibrant startup located in Berlin.

My daily challenges revolved around developing and refining Natural Language Processing (NLP) models using BERT and enhancing Computer Vision models with Faster-RCNN. My primary goal was to boost the accuracy of existing models, and I was eager to explore these technologies.

A Shift in Perspective

During casual discussions about my professional growth, my supervisor consistently highlighted a point that initially left me quite confused.

He emphasized the importance of focusing on the production aspect of model development, rather than solely on experimentation.

At the time, his remarks baffled me. I believed that the responsibility of a data scientist was limited to understanding business needs, performing statistical analyses, and identifying optimal models. What more could there be?

Connecting to Production

This was a significant revelation for me. The idea of “deploying the model into production” was unfamiliar territory. I had not encountered this topic in my academic studies, nor was it part of my 2022 skill set.

As weeks progressed, the concept gradually became clearer.

I began to understand my supervisor's emphasis and recognized the necessity to enhance my Data Science skills.

It became evident that it wasn't only about identifying the best model or conducting statistical analyses; it was crucial to ensure that these models could be successfully integrated into the company’s systems and workflows.

This integration was the key to transforming my experimental work into tangible solutions that could generate business value.

While my internship primarily focused on experimentation in NLP and Computer Vision, this guidance from my supervisor significantly impacted my career trajectory.

Greetings!

I’m Sara, a Data Scientist at EDP, holding a Master’s degree in Physics, I transitioned into the dynamic field of Data Science.

I write about data science, AI, and career guidance in these domains. If you want to keep reading, be sure to subscribe and follow me!

While formal education provides essential skills, it often fails to equip students with the practical knowledge necessary in the corporate world.

Last month, while assisting someone in selecting courses for their data science master’s program, I noticed a glaring absence of courses focused on model production or even introductory topics in this area.

Deploying models is a facet of a broader subject known as Machine Learning Operations (MLOps). However, MLOps is just one example; there are several other critical skills that traditional education overlooks, which I will explore in this article.

For instance, cloud computing knowledge is increasingly vital. As more organizations transition their operations to the cloud, proficiency in platforms like AWS, Azure, or Google Cloud can greatly enhance your capacity to deploy and scale machine learning models.

Learning to utilize the various tools and services offered by these platforms can optimize your workflow and minimize infrastructure costs.

You certainly won’t need to master every skill! Your requirements will vary depending on your goals and preferences.

However, by broadening your skill set beyond traditional teachings, you can emerge as a more adaptable and valuable data scientist.

This article will outline these skills and more, offering a comprehensive guide to what you need to excel in 2024 and beyond!

Why Should You Care?

Evolving Skills

A few years back, essential skills for a Data Scientist included Python/R, machine learning, SQL, data visualization, and statistics. However, Data Science is a constantly evolving field!

The skills sought in 2017 are quite different from those necessary in 2024, a natural evolution driven by technological advancements and shifts in the job market.

Impact of AI

The recent surge in Generative Artificial Intelligence (GenAI) and large language models (LLMs) underscores the growing demand for AI skills.

Does this imply that AI skills will be more sought after? Absolutely.

> While deep expertise in **NLP or LLMs* (crucial subsets of AI) may not be required for a data science role, a solid understanding of AI systems, their business implications, and their potential to create value is becoming essential for any data-driven career.*

The Expanding Role of Data Scientists

#### Overlap of Roles

> But what if I don’t aspire to be a Machine Learning Engineer? Does this still pertain to me?

That's a valid inquiry. Yet, many current job postings for Data Scientists include requirements that overlap with Machine Learning engineering skills.

While some organizations clearly distinguish between Data Scientists, ML Engineers, and MLOps engineers, there is often significant overlap.

And this makes perfect sense! It benefits companies to have employees who possess a foundational understanding of multiple domains, fostering effective communication and collaboration among various roles.

#### Example of a Job Posting

In the image below, you can see a recent Data Scientist job posting I captured from LinkedIn (May 2024). This role is heavily centered on NLP and GenAI, despite being classified as a Data Science position.

In this article, I will delve into some of the key skills highlighted in that job post.

Furthermore, I’ve observed an increasing trend of ML Engineer and AI Engineer positions. It is anticipated that demand for AI and ML roles will rise by 40% from 2023 to 2027!

This change is reflected in the job market, where companies are capitalizing on recent AI advancements by hiring more ML/AI Engineers.

Individuals with backgrounds in Data Science who are eager to learn new skills can easily adapt to shifts in the job market!

In this article, I will outline five essential skills you need to remain relevant in today’s employment landscape. These skills will not only elevate your standing but also offer greater flexibility in accessing job opportunities!

For each skill, I will discuss:

• The necessity of the skill;
• Its daily applications by Data Scientists;
• Real-world examples of how Data Scientists utilize the skill;
• Effective learning methods.

Let’s get started!

Deep Learning

Why You Need this Skill Now

Deep learning has transformed areas such as image and speech recognition, autonomous driving, and predictive analytics.

Its capacity to process vast amounts of data and uncover complex patterns makes it essential for contemporary data scientists.

Moreover, deep learning leads many innovative technologies, including generative adversarial networks (GANs) and reinforcement learning, pushing the boundaries of artificial intelligence.

These advancements are not only reshaping current industries but also creating entirely new sectors, spanning from creative fields to autonomous systems.

As various industries adopt deep learning technologies, the demand for professionals capable of designing and implementing these intricate models will surge.

Acquiring knowledge in deep learning is a vital step toward mastering the subsequent skills discussed in this article.

> Understanding the fundamentals of deep learning lays a solid foundation for comprehending and executing advanced ML and AI techniques.

How do Data Scientists Apply this Skill?

Data scientists employ deep learning techniques in various ways to create sophisticated models for tasks such as image classification, forecasting, natural language processing, and anomaly detection.

#### Real-World Example

Imagine you work for a financial institution.

Deep learning models can be utilized to identify fraudulent transactions in real-time. By analyzing transaction patterns and spotting anomalies, these models assist financial organizations in preventing fraud and safeguarding customer assets.

How to Learn this Skill?

Before delving into deep learning, ensure you have a fundamental grasp of mathematical concepts (linear algebra, calculus, probability, and statistics).

Start with the core principles of deep learning: neural networks and backpropagation.

Gain hands-on experience with leading deep learning frameworks (TensorFlow, PyTorch).

Next, progress to more complex models (CNNs and their architectures).

Understand models designed for sequence data (RNNs, LSTMs...).

Once you’re adept with CNNs and RNNs, you can explore more advanced topics: Generative Adversarial Networks (GANs) and Reinforcement Learning.

AI and Machine Learning Exploration

Why You Need this Skill Now

AI and machine learning are currently revolutionizing industries.

The rise in AI technologies, particularly NLP, LLMs, and GenAI, has rendered AI skills increasingly valuable.

Understanding how these technologies function and their potential applications is vital for maximizing their business and tech impact.

AI will continue to manifest in areas such as predictive analytics, automation, anomaly detection, chatbots, and intelligent systems.

Organizations are heavily investing in AI technologies to discover new avenues for delivering value to their customers, which explains the increased demand for skilled professionals in these areas.

How do Data Scientists Apply this Skill?

Data Scientists have myriad opportunities to leverage AI today!

In 2024, they employ AI to enhance various facets of their work.

Automated Machine Learning (AutoML) simplifies the model development process, while advanced NLP and computer vision applications yield deeper insights into text and image data, automating several tasks.

In terms of time-series analysis, advanced AI algorithms are utilized to predict future trends, thereby improving forecasting accuracy.

AI-driven data preprocessing and integration enhance data quality and accessibility.

Explainable AI (XAI) boosts model transparency and aids understanding of decision-making processes.

AI models are also instrumental in identifying unusual patterns (anomalies) that deviate from expected behavior, proving beneficial in fraud detection, network security, and predictive maintenance.

#### Real-World Examples

One of the most apparent applications of AI and machine learning since the AI boom is in the development of chatbots and virtual assistants.

These AI-driven tools utilize NLP to comprehend and respond to customer inquiries, offering support and information around the clock.

For instance, numerous companies have integrated AI chatbots into their customer service operations to address routine queries.

Additionally, AI is being utilized in innovative ways, such as personalized recommendation systems, predictive maintenance in manufacturing, and smart healthcare systems that analyze patient data to propose treatments.

How to Learn this Skill?

Embarking on the journey to learn LLMs and GenAI can seem overwhelming.

I recommend starting with the basics.

You don’t need to master every model, framework, or skill.

If you wish to specialize in a particular area (e.g., becoming an NLP Engineer), feel free to pursue that path.

However, if your aim is to enhance your skills to remain relevant in the job market or to satisfy your curiosity, having sufficient knowledge to execute tasks should suffice in the initial phase!

To kickstart your AI journey and advanced machine learning, begin with foundational courses that cover basic concepts, gradually building your expertise.

Assuming you have a solid grasp of common ML models (Linear and Logistic Regression, Random Forests, clustering algorithms, etc.), you can delve into more complex topics.

Explore NLP, a critical AI component focusing on human-computer interaction. Begin with text processing, text classification, and word embeddings.

Ensure you have a robust understanding of deep learning (which is why I highlighted it as the first skill in this article).

Then, you can dive deep into LLMs. Familiarize yourself with transformers, BERT, and GPTs, and learn about training and fine-tuning these models.

Since the AI boom, the availability of free resources online covering this topic has skyrocketed!

Cloud Computing Basics

Why You Need this Skill Now

Understanding how to utilize cloud platforms allows data scientists to access powerful infrastructure, reducing costs and simplifying data storage and processing.

Cloud computing provides scalable and flexible resources, ideal for handling large datasets and intricate calculations.

Additionally, it enhances collaboration and accelerates model deployment, enabling you to concentrate more on analysis and less on managing hardware.

How do Data Scientists Apply this Skill?

Data scientists utilize cloud platforms to:

• Store data: Centralize all data without concerns about storage limits.
• Process data: Use robust tools for rapid data cleaning and analysis.
• Experiment with models: Easily test various machine learning models.
• Track experiments: Maintain records of all experiments and their outcomes.
• Deploy models: Seamlessly launch models into practical applications.
• Collaborate: Work collectively with a team on shared data and models.
• Automate workflows: Establish automatic processes for repetitive tasks.

Cloud services promote collaboration, scalability, and efficient resource management, making them indispensable in modern data science initiatives!

Real-World Example

A data scientist at a healthcare startup might leverage a cloud platform to build and deploy a machine learning model that analyzes medical imaging data, detecting early signs of diabetes, thus facilitating quicker and more accurate patient diagnoses.

How to Learn this Skill?

Begin by learning one cloud platform. You don’t need to be proficient in all!

From my experience, mastering one platform equips you to transition to another easily. Companies prioritize practical experience with any single platform.

Personally, I began with Microsoft Azure. The three most prevalent platforms are Amazon Web Services (AWS), Azure, and Google Cloud Platform (GCP).

Machine Learning Model Deployment

Why You Need this Skill Now

Creating a machine learning model is merely the initial step. Deploying these models in production environments is essential for delivering real-time predictions and generating business value.

Being proficient in deployment guarantees that models can scale, integrate smoothly with existing systems, and function effectively across various scenarios.

Is this skill requisite for all data science roles? No. However, as previously mentioned, there's often an overlap with machine learning engineer skills, making it beneficial to understand the fundamentals for efficient communication with other engineers.

Furthermore, if data scientists comprehend the prerequisites for deployment and the available resources, it aids in selecting the most suitable model for the project.

How do Data Scientists Apply this Skill?

Data Scientists frequently work closely with machine learning engineers to package, optimize, and integrate models into production systems.

They ensure that models are scalable, secure, and maintainable by establishing APIs, monitoring systems, and CI/CD pipelines.

It’s crucial for Data Scientists to possess machine learning model deployment skills, as it enables them to transform their models into actionable, real-world applications.

It is vital that their insights and predictions can effectively drive business value and support decision-making processes.

#### Real-World Example

You may be familiar with e-commerce platforms.

Companies deploy recommendation engines to suggest products to customers based on their browsing and purchasing histories.

These models are integrated into the website’s backend to deliver real-time recommendations, enhancing the user shopping experience.

How to Learn this Skill?

After mastering the fundamentals of machine learning, begin familiarizing yourself with cloud platforms such as AWS, GCP, or Azure.

You should also study various deployment tools like Docker or Kubernetes (start with Docker, as it is more beginner-friendly and widely utilized).

Additionally, familiarize yourself with good MLOps practices to manage the lifecycle of machine learning models (continuous integration, continuous deployment (CI/CD), and monitoring).

Big Data

Why You Need this Skill Now

“Today, all data is big data,” or so the saying goes.

While this isn’t always accurate (I’m currently engaged in a project with just 150 rows of data...), it generally holds.

With the exponential growth of data, possessing big data skills is imperative.

These abilities empower you to efficiently process and analyze massive datasets, helping you draw valuable insights and make informed decisions.

Big data skills encompass a variety of proficiencies, including knowledge of programming languages (like Python, Java, and SQL), an understanding of data structures and algorithms, and familiarity with big data processing frameworks such as Apache Hadoop and Spark.

These skills are vital for managing datasets that exceed the capabilities of conventional data processing tools.

How do Data Scientists Apply this Skill?

Data scientists utilize big data skills to manage, process, and analyze extensive volumes of data using tools like Hadoop, Spark, and distributed databases, allowing them to unearth patterns and insights that might remain hidden in smaller datasets.

#### Real-World Example

A data scientist at an environmental research organization employs Apache Spark to process and analyze satellite imagery for remote sensing applications.

Every day, they handle terabytes of high-resolution images from multiple satellites, using Spark to efficiently clean, preprocess, and store this vast amount of data.

They apply machine learning algorithms to identify changes in land use, monitor deforestation, and assess the health of agricultural crops.

For instance, by analyzing spectral data from satellite images, they can pinpoint areas affected by drought or disease.

How to Learn this Skill?

Begin by mastering one big data platform or tool, such as Apache Spark or Hadoop. Engage in online courses, work on real-world projects, and practice managing large datasets to enhance your proficiency.

Final Thoughts and Key Takeaways

Why is it essential to acquire new data science skills in 2024?

The answer is straightforward: the field of data science is rapidly evolving.

Keeping abreast of new skills is crucial for maintaining a competitive advantage and continuing to add value to organizations.

As explored throughout this article, the role of a data scientist is expanding, along with the necessary skills!

The World Economic Forum’s Future of Jobs Report 2023 indicates that AI and big data skills are among the top in-demand skills and will continue to grow in importance through 2027.

According to this report, 60% of workers will need training in these areas as businesses rapidly embrace advanced technologies.

This implies that investing time in learning these skills now will yield significant returns in the future.

Reflecting on my internship at the Berlin startup, I recognize how invaluable that experience was.

It taught me that being a data scientist transcends the excitement of discovery and analysis; it also involves ensuring that those discoveries can be practically applied.

This lesson remains with me, constantly reminding me to balance experimentation with production.

Thus, I urge you to begin developing these skills today!

Whether it’s deep learning, AI and ML, cloud computing, model deployment, or big data, each of these domains is critical.

Not only will they enhance your versatility, but they will also create new opportunities and enable you to make a greater impact in your field.

Remember, the journey of learning and growth is ongoing. Embrace these new skills and continually push the limits of what’s achievable in data science. Your future self will be grateful!

Thank you for reading!

If you found value in this post, I’d appreciate your support with a clap! You’re also welcome to follow me on Medium or LinkedIn for similar articles!

Curious about how I transitioned from Physics to Data Science? Check out the article below!

<div class="link-block">

<div>

<h2>How to Transition from Physics to Data Science: A Comprehensive Guide</h2>

<div><h3>Advice from a Physics Master’s Graduate turned Data Scientist</h3></div>

<div><p>towardsdatascience.com</p></div>

</div>

<div>

</div>

</div>

Do you work with time-series data? Then you must check out the article below!

<div class="link-block">

<div>

<h2>The Ultimate Guide to Finding Outliers in Your Time-Series Data (Part 1)</h2>

<div><h3>Effective statistical methods and tools for outlier detection in time-series analysis</h3></div>

<div><p>towardsdatascience.com</p></div>

</div>

<div>

</div>

</div>

My name is Sara Nóbrega, and I am a Data Scientist with a background in Physics and Astrophysics. I’m passionate about AI, MLOps, Smart Cities, Sustainability, Cosmology, and Human Rights.

References:

• The state of AI in 2023: Generative AI’s breakout year | McKinsey
• Jobs In Data — Machine Learning Engineer vs Data Scientist — Salary Gap (jobs-in-data.com)
• Data Scientist Job Market 2024: Analysis, Trends, Opportunities | 365 Data Science
• Machine Learning Engineer Job Outlook 2023: Research on 1,000+ Job Postings | 365 Data Science
• The 10 Most Important Data Science Skills in 2023 — HackerRank Blog
• The AI Talent Rush: 10 In-Demand AI Jobs for 2024 (onwardsearch.com)
• AI investment forecast to approach $200 billion globally by 2025 (goldmansachs.com)
• The Future of Jobs Report 2023 | World Economic Forum (weforum.org)