Inside the new AI jobs: How companies are hiring and what they’re paying
From data annotators to AI security researchers, companies are already filling roles that barely existed two years ago.
The predictions that tens of millions of jobs will be lost due to the introduction of AI technologies in the next five years are starting to look more tangible than ever, in light of waves of layoffs around the world and in Israel. The numbers vary according to different reports, but the picture that emerges about the future is that while tens of millions of jobs will disappear, new jobs will also be created thanks to technology, so that, in the end, there will be more new jobs than those lost. According to reports from the World Economic Forum, if we subtract the jobs that will be lost from those that will be created, we will witness the creation of about 14 million new jobs worldwide. These roles are not somewhere in the distant future, they are already starting to be created here and now, and you can see them being filled in companies and in job advertisements.
"The labor market is entering a new era where artificial intelligence is not replacing workers, it is expanding the possibilities for them. We are seeing more new roles being born around AI, especially hybrid professions, which combine deep technological understanding with business insight and the ability to work hand in hand with the customer," says Einat Frish, Director of Recruiting - EMEA Tech, at Salesforce. "In my opinion, this is a tremendous opportunity for employees to bring to life a wider range of skills, and to build a career that evolves along with technology. Organizations that adopt AI will benefit not only from efficiency, but also from a more diverse and skilled workforce, and this is what will give them the competitive advantage in the coming years."
Unicorn Gong is also seeing rapid growth in new AI roles. "The shift from focusing on training models to adding agents and advanced capabilities like RAG or Knowledge Graph requires skills that were not required in the past. Therefore, roles such as AI Engineer, who are responsible for quickly setting up agent infrastructure and connecting it to the product, and Prompt Engineer, who focuses on creating and optimizing prompts for features based on generative models, have been created," says Danit Berger, NLP Manager at Gong, a leader in the field of Revenue AI.
So, what new roles have been created and already exist? What do they actually do, how much do they earn, and what training is required? These are the hottest AI professions today, some of which are even entry-level roles or ones that don't require any technological training:
1. Data Annotator: The Algorithm's "Private Tutor"
The role: Digital Accessibility or Data Annotator, a data analyst (digital accessibility). Essentially, the algorithm's "private tutor." If in the past QA professionals were required to check code, today people are required to check that the model understands the world correctly and bridges the gap between raw code and human reality.
"At Evinced, we solve the accessibility problem at scale, directly within the source code. Generic AI tools tend to hallucinate and produce inaccessible code. This role is the human factor that creates boundaries for our AI. Trainers teach the model the complex logic of digital accessibility, so our tools can guide developers in real time, from the design stage to automatic correction," says Yossi Synett, Chief Scientist at Evinced.
What is actually done in the role?
The daily work combines quality control and education, transforming human knowledge into a digital brain that prevents malfunctions before they occur:
Building the Ground Truth: Creating and correcting high-quality datasets used to train and evaluate models, so they can identify accessibility failures that normal automation misses. This is done through annotation and precise classification of patterns in source code according to a structured table of rules, preventing AI from producing inaccessible code.
"The field of data labeling has existed for about a decade (mainly for image recognition), but the role in its current version, as a structured digital accessibility data trainer, has exploded in the last two years. Since the introduction of ChatGPT and large models, the need has gone from acute to critical. Models can answer, but to be useful for business and not produce nonsense, they need this human layer being built right now. We realized we need human trainers to convey subtle nuances of user experience and digital accessibility, which no machine can yet learn on its own," says Synett.
Job Requirements: Sharp logical thinking, attention to detail, and basic technical orientation; no development experience required.
Training: Almost no formal university courses exist, presenting an opportunity. Evinced trains employees in digital accessibility principles and its proprietary methodology for identifying components. Rapid training integrates employees into the startup’s core operations.
Salary: 65–100 NIS per hour ($20-$30).
2. Prompt Engineer: "The Whisperer to Artificial Intelligence"
The Role: Prompt Engineers develop, improve, and manage features based on language models (LLMs). They work with research and data departments to build intelligent AI capabilities.
What is actually done in the role? Designing experiments, building prompt workflows, creating evaluation methodologies, and implementing them in products using complex architectures. Work is collaborative with research, development, and product teams.
Job Requirements: Analytical and product thinking, deep familiarity with LLM behavior, and basic development knowledge. Candidates may come from analytics, development, or product roles with a technological orientation.
Training: Available internally at Gong or via online courses from OpenAI or DeepLearning AI.
Salary: Varies depending on background and role definition.
3. AI Security Researcher: Identifying Weaknesses
The Role: Identifies weaknesses in AI models and develops defenses against attacks and information leaks.
What is actually done in the role? Examining model vulnerabilities, finding loopholes, developing defensive tools, assessing model resilience, and collaborating with development teams.
"This role is relatively new, existing only for about one to two years, born from the pace at which language models entered real products," says Johnathan Azaria, ML Platform Team Lead at Imperva.
Job Requirements: Data scientists with cyber backgrounds or cyber researchers with knowledge of language models; research ability and familiarity with AI attacks.
Training: No formal track; most learn on the job.
Salary: 40,000-60,000 NIS ($12,500-$18,700) depending on experience.
4. FDE (Forward Deployed Engineer: The Field Engineer)
The role: A hybrid engineer, consultant, and implementer who ensures AI projects move from pilot to live product at customer sites.
What do you do in the role? Understand organizational needs, implement tailored AI solutions, serve as an organizational interpreter between business and development, and provide feedback to the product team.
Job Requirements: Technical AI experience (e.g., RAG, Front/Back-End development) plus strategic understanding (prompt engineering, data analytics). Typically experienced software engineers with a learning mindset.
Training: Internal company programs.
Salary: Comparable to software engineers.
5. Fullstack AI Engineer: Holistic Developer
The role: End-to-end AI product development, integrating AI from idea to production.
What is actually done in the role? At April, integrates AI capabilities into tax-related products, builds evaluation systems, organizes data, develops human-machine interactions, and performs prompt engineering.
Job Requirements: Six+ years full-stack development experience, at least one year in AI product development.
Training: Internal training only.
Salary: 30,000–50,000 NIS ($9,300-$15,500)
6. Director of AI Content & Innovation: The Bridge Between Content and Technology
The role: Leads innovation at the intersection of technology, content, and product.
What is actually done in the role? Examines new AI technologies, plans user journeys, and educates customers on tool usage.
Job Requirements: Multidisciplinary skills in AI, coding, content creation, and product management; mental flexibility and fast learning required.
Training: Practical, on-the-job experience.
7. AI Optimization Specialist (AIO)
The Role: Optimizes content and organizational knowledge for LLMs, similar to SEO for search engines.
What is actually done in the role? Optimizes information retrieval and accuracy, runs experiments, collaborates with teams, and monitors AI performance.
Job Requirements: Understanding of LLM behavior, prompt design, content structuring, analytical thinking; technical experience is optional.
Training: Developing programs in prompt engineering and AI content optimization; internal company training.
Salary: 20,000-25,000 NIS ($6,200-$7,800).
8. AI Developer: Next-Generation Developer
The role: Develops advanced GenAI solutions, including chatbots, RAG infrastructures, and agent-based applications.
What is actually done in the role? Works with data and cloud teams to deploy multi-agent architectures and modular AI services in Python.
Job Requirements: 1-3 years LLM production experience, Python and cloud proficiency, vector database experience.
Training: On-the-job learning.
Salary: Varies by experience.
9. AI Director / Head of AI Transformation
The role: Leads AI adoption across departments, driving cultural and organizational change.
What is actually done in the role? Reexamines processes, implements automation, and aligns AI tools with business needs.
Job Requirements: Strong technology-business hybrid skills, curiosity, fast learning.
Training: Practical experience; no formal academic track.
Salary: Comparable to development managers.
10. Product Engineering: Product Engineer Integrating AI
The role: Integrates AI tools into products from research and prototyping to release, accelerating development and productivity.
What is actually done in the role? Uses AI tools throughout the product lifecycle, enabling advanced development at early stages.
Job Requirements: Product background with technical understanding, strategic thinking, systemic vision, rapid learning, customer management.
Training: Internal company training tailored to proprietary tools and processes.
Salary: Based on experience and technical background, aligned with market ranges.
