How to hire for AI proficiency (without getting fooled by buzzwords)

What AI proficiency actually means

Before we dive into steps for hiring skilled AI software engineers, you should understand what AI proficiency really means. Talented AI designers and managers have various technical and non-technical skills, including:

• Foundational understanding of AI: Knowledge about the core workings of AI and machine learning techniques, including unsupervised learning, reinforcement learning, and natural language processing.

• Programming language and framework fluency: Experience using key programming languages and packages for AI development, including Python, TensorFlow, PyTorch, scikit-learn, LangChain, and HuggingFace.

• Data handling experience: Proficiency at preprocessing and validating data, managing SQL and NoSQL databases, and conducting statistical analysis, data visualization, and data labeling.

• Model evaluation skills: Experience performing model cross-validation, developing and benchmarking performance metrics, and fine-tuning pretrained models.

• Deployment experience: Proficiency in machine learning operations (MLOps), containerization using tools like Kubernetes and Apache Airflow, version control, and continuous development.

• AI ethics and explainability: Experience building interpretability into machine learning models and using explainable AI techniques like Local Interpretable Model-agnostic Explanations (LIME) and Shapley Additive Explanations (SHAP).

• Problem-solving and adaptability: Comfort working on challenging projects, including the ability to think critically and ensure AI projects meet real business needs.

Our guide to in-demand AI skills covers each of these skills in more depth so you can know exactly what to look for in AI candidates.

Why hiring for AI skills is so challenging right now

If hiring for AI roles seems especially difficult right now, that’s because it is. In a recent survey of corporate leaders, 62% said a lack of employees with the necessary AI skills is holding them back from reaching their AI deployment goals. 

Hiring managers are facing a few problems. The biggest? The pool of talented AI applicants simply isn’t large enough. According to Indeed, candidates with AI skills command salaries up to 47% higher than their peers, indicating a significant shortage in qualified applicants. 

This is in part because demand for machine learning engineers is surging right now. It’s also because many candidates have only academic knowledge and no hands-on experience developing AI models. That isn’t good enough when your company needs a production-ready model. Many candidates can train AI language models, but few can deploy them reliably, handle messy data, or work effectively in cross-functional AI teams.

On top of all this, hiring managers are being inundated with resumes that overstate applicants’ AI experience. That makes it even harder to screen applicants and make the best hire.

5 steps to hire for AI proficiency

The best way to overcome the challenges around hiring AI experts is to develop a skills-based hiring process that’s laser-focused on AI proficiency. Here’s how you can hire for AI proficiency in five steps.

Step 1: Define the role clearly

The first step in attracting talented AI engineers to your job is to clearly define the role you’re hiring for. This is important because AI-related roles can vary widely. For example, different AI roles could include:

• AI engineers focused on building, scaling, and maintaining large language models and other AI products

• Data scientists focused on cross-functional collaboration and designing new product features

• Data engineers focused on structuring and labeling data for AI model training

• Reinforcement learning specialists focused on model training and optimization

• Software developers (including game devs) focused on incorporating AI models into existing products

Each of these roles requires distinct skills and specializations, so it’s important to be specific about which type of AI employee you need. In addition, because the AI field is so broad, talented applicants want to know what they’ll be responsible for in a role and often won’t apply to vague job postings.

You can further enhance your AI job posting by being specific about:

• The role’s business context, including what team or department a new hire will be part of

• Which types of datasets the role involves working with

• Which specific programming languages, frameworks, and AI tools will be used in the role

• Which products and deliverables the new hire will be responsible for creating

Check out our guides to how to write computer vision engineer job descriptions and generative AI engineer job descriptions for help creating the perfect job posting.

Step 2: Source from the right pools

Once you have a strong job description, it’s important to get it in front of talented candidates. While traditional online job boards are a good place to start, you’ll increase your chances of attracting strong applicants by advertising your role in places where AI experts are most likely to be working.

For example, be sure to post your role on AI-specific job boards. Another good approach is to reach out directly to qualified AI developers you find on GitHub or Kaggle to invite them to apply to your job.

Other ways to find prospective applicants include:

• Looking at AI research paper authors on Arxiv

• Searching for experts at AI research labs

• Partnering with AI bootcamps to hire talented students

• Networking at AI developer conferences

Step 3: Assess candidates with practical skills tests

Once you have a quality pool of applicants for your role, the next step is to assess candidates’ AI proficiency using practical tests for core AI skills.

Skills tests are much more effective at identifying quality candidates than other screening methods.

Jeff Caiden, CEO of third-party logistics firm Capacity LLC, says, “When it comes to hiring, we don’t rely only on resumes or standard interviews. We create real-world problem scenarios tied to our operations. We want to see how someone thinks, how they explain their approach, and whether they can handle ambiguity. You can tell a lot about a person by watching how they work through an operational challenge that doesn’t have one clear answer.”

TestGorilla makes it easy to assess candidates’ hands-on skills with our AI test and skill-specific tests for:

• Data wrangling

• Machine learning

• Neural networks

• DevOps

We also have coding skills tests for programming languages like Python and R, plus other programming skills tests for frameworks and libraries like Apache Spark, TensorFlow, scikit-learn, and more. Our cognitive ability tests enable you to assess a candidate’s critical thinking and problem-solving skills. 

You can combine several of these tests to create a complete assessment that’s tailored to the demands of your role. You can also add tests for soft skills like communication and collaboration.

The result is an objective, efficient screen that enables you to quickly identify the most talented applicants and gain insights into candidates’ strengths and weaknesses.

Step 4: Screen candidates with the right signals

After conducting skills-based testing, you can dig deeper into applicants by looking at their past work.

Check applicants’ GitHub profiles, Colab notebooks, and other public codebases to see what code they’ve developed and what projects they’ve been involved in. Also, be sure to look for blog posts or research papers that candidates have authored, which can provide deeper insight into their area of expertise.

This is also an opportunity to look at applicants’ resumes if you choose. When reading resumes, filter out buzzwords like “ChatGPT architect” or “machine learning expert” and don’t rely on past positions at major tech companies as a signal of AI proficiency. Instead, focus on the projects that candidates have worked on and the specific technical skills they bring to the table. 

This information can help you prepare for candidate interviews and identify which applicants have proven, hands-on experience working with AI.

Step 5: Interview for business context + communication

Finally, you can use interview questions to evaluate AI proficiency. Candidates should be able to answer questions about AI fundamentals, as well as more difficult questions about how they approach trade-offs in AI models – like precision vs. recall and latency vs. accuracy. They should also have a clear grasp on the impacts their work has on products, internal stakeholders, and customers – and be prepared to work in ways that best achieve your company’s goals.

You can also ask about an applicant’s previous experience working in cross-functional teams or solving complex problems alongside peers. These questions can shed light on a candidate’s collaboration and communication skills.

Where possible, use scenario-based questions such as: “You built a high-performing model that isn’t getting adopted. What do you do?” These types of questions encourage candidates to discuss their approach to tricky situations that might involve both technical and non-technical solutions. 

4 things to avoid when hiring for AI proficiency

The steps above can help you develop an effective hiring process for critical AI roles. However, it’s important to avoid a few common mistakes that can hold you back from hiring top candidates.

1. Forgetting AI ethics

Designing for ethical AI is increasingly important for complying with regulations and mitigating reputational risk. Be sure to ask candidates how they approach AI ethics and what experience they have dealing with model bias.

2. Not thinking about your company’s existing tools

It’s easy to be wowed by an AI expert’s knowledge of so many programming languages and frameworks. But think carefully about which languages, frameworks, and other tools your company is currently using, and make sure an applicant has deep experience using those tools – or the ability to quickly learn them.

3. Asking for too much work

It’s acceptable to ask candidates to take a 30–60 minute skills assessment and sit for an interview before hiring. But be careful about asking for too much work, like large coding demonstrations, as this can discourage skilled AI engineers from continuing through your hiring process.

4. Ignoring culture add

A candidate might have all the technical skills required for a role, but if they won’t fit with – or better yet, enhance – your company’s culture, think twice about hiring them. Culture fit is crucial for strong collaborations and employee morale.

Hire top AI professionals today

Finding talented candidates for AI roles can be challenging, but tailoring your hiring process to focus on AI proficiency can ensure you make the best hire every time. Be sure to clearly define the role you’re hiring for and advertise your job in places AI experts are likely to see it. Then, screen applicants using a combination of role-specific skills assessments, deep candidate research, and tailored interview questions.

TestGorilla makes the screening process easy with our AI-specific skills tests and holistic, multi-measure testing approach. Book a demo today to start hiring for AI proficiency.