Careers in AI: the 8 paths into the industry, and who's hiring

Careers in AI now split into eight paths: machine learning, LLM and GenAI, MLOps and infrastructure, research, data science, deployment and operations, product and design, and marketing. Five are engineering or science jobs. The other three face customers and the business, and they hold about 4 in 10 of the AI jobs we list.

On Tokn, as of 12 October 2026, we list about 770 open AI roles at 79 companies in Europe, from company career pages. This guide uses that data to show what each path involves, who it suits and who is hiring. We track 86 companies, with a tilt towards AI-native firms in France and the UK, so the figures describe our board, not the whole European market.

In short

• The two largest AI career paths on Tokn are machine learning (171 jobs) and AI deployment and operations (158).
• About 1 in 5 jobs on Tokn is a business-facing role with no engineering in the title. Most deployment and operations roles are still engineering jobs.
• The market skews senior. Juniors have the best odds in machine learning and data science.
• Five companies hold a third of all openings. About one job in six is fully remote; London and Paris lead the rest.

Careers in AI at a glance: the 8 paths

These eight paths cover every AI job on Tokn. Three are business-facing, though the biggest of them is mostly staffed by engineers.

"Fully remote" is the share of all jobs in the path flagged as fully remote. About 40% of postings don't state a work policy, so treat these shares as a floor.

Technical AI career paths

Python appears in most postings on these five paths. What changes is the rest of the stack and how much experience you need.

Machine learning engineer and AI engineer

The largest path on Tokn, with 171 jobs, is about building models and AI features into real products. Titles have shifted: across the board, "applied AI" appears in 62 job titles and "AI engineer" in 56, against 33 for "machine learning engineer".

• Suits you if: you are a software engineer or data scientist who wants to ship models, not only study them.
• Typical titles: Applied AI Engineer, Enterprise (Anthropic); AI Engineer (Granola, Legora); Senior AI Engineer (Doctolib, Veepee).
• Skills postings ask for: Python (58%), TypeScript (23%), AWS (20%), PyTorch (19%), RAG (17%).
• Who's hiring, and where: Anthropic, OpenAI, Wayve and Celonis lead. Only 13% of roles are fully remote; Paris and London lead, with Madrid and Munich behind.
• Way in: more junior roles (23) and internships (11) than any other path.

Browse all machine learning jobs.

LLM and GenAI engineer

LLM engineers build products on top of large language models: agents, assistants, retrieval and the interfaces around them. OpenAI sums up one such role as helping "bring Codex agents from impressive demos to dependable tools".

• Suits you if: you ship product code and want to own LLM features end to end.
• Typical titles: Applied AI Engineer, Codex Core Agent (OpenAI); Principal AI Engineer, Agents and Context (Elastic).
• Skills postings ask for: Python (66%), TypeScript (42%), Kubernetes (31%), AWS (22%): the most full-stack mix.
• Who's hiring, and where: 76 jobs spread across many companies, led by ElevenLabs, Elastic and Mistral AI. It is the most remote-friendly path: 20 jobs (26%) are fully remote.
• Way in: it leans senior, with only 6 junior roles.

For a closer look, browse the LLM and GenAI jobs open right now.

MLOps and AI infrastructure engineer

These engineers run the systems that models train and serve on, which ElevenLabs describes as everything "from the GPU clusters that train and serve our models to the cloud platform our product teams build on". Almost nobody puts "MLOps" in a title any more (2 jobs on Tokn): search for platform or infrastructure engineer.

• Suits you if: you prefer distributed systems and reliability to modelling.
• Typical titles: Infrastructure Engineer (Dataiku, ElevenLabs); AI Platform Engineer (Supabase); AI Infrastructure Systems Engineer (Together AI).
• Skills postings ask for: Kubernetes (68%), Python (58%), AWS (44%), GCP (40%), Terraform (38%).
• Who's hiring, and where: 91 jobs, led by Mistral AI, Wayve and Together AI. 20% are fully remote; London and Paris lead.
• Way in: the most senior path. Of the 58 roles with a stated level, 47 are senior or lead and 6 are junior.

Browse MLOps and AI infrastructure jobs.

Research scientist and research engineer

Research teams train and improve models, and more and more they ship that work themselves. On Tokn, "research engineer" titles outnumber "research scientist" more than 3 to 1 (44 vs 13).

• Suits you if: you are strong in maths and deep learning and enjoy reading papers.
• Typical titles: AI Scientist (Mistral AI); AI Research Engineer, Foundation Models (Helsing); Research Engineer / Scientist, Alignment (Anthropic, London).
• Skills postings ask for: Python (67%), PyTorch (52%), JAX (26%), Kubernetes (22%), Rust (22%), then C++ and CUDA.
• Who's hiring, and where: 91 jobs, led by Mistral AI, Anthropic and Helsing. 20 jobs (22%) are fully remote, the second-highest share after LLM & GenAI; London and Paris lead.
• Way in: most research postings don't state a level, so we can't give a reliable seniority picture.

Browse AI research jobs.

Data scientist

Data scientists analyse data and build models that inform business decisions; analytics engineers, filed in the same path on Tokn, build the pipelines behind them. It is the smallest path (49 jobs).

• Suits you if: you like statistics and SQL, and explaining numbers to people who don't.
• Typical titles: Data Scientist (Artefact, Lovable, Mistral AI); Analytics Engineer (ElevenLabs, Modal); Data Scientist, AI Safety (ElevenLabs).
• Skills postings ask for: Python (75%), SQL (69%), BigQuery (20%), dbt (20%), Azure (16%).
• Who's hiring, and where: Artefact, Ekimetrics, Lovable and Shift Technology lead, making it the path most driven by consultancies. It is also the least remote: only 5 jobs (10%) are fully remote, and London and Paris lead.
• Way in: the highest junior share of any path (8 of 33 roles with a stated level, a small sample).

Browse data science jobs.

Business-facing AI careers: can you work in AI without coding?

Yes, but fewer roles are code-free than the category names suggest. The three business-facing paths hold 296 jobs, or 38% of the board. But about 4 in 10 of their titles say engineer, developer, scientist or architect, leaving about 1 in 5 jobs on the whole board as business-facing roles with no engineering in the title.

Many of those roles still expect technical fluency. And the pattern isn't only European: in the US, Lightcast's summary of the Stanford AI Index 2026 notes that "a majority of AI job postings come outside IT roles".

AI deployment, solutions and forward deployed roles

AI Ops & Automation is the second-largest path (158 jobs) and the most misunderstood: nearly two-thirds of its titles are engineering titles. At one end, the forward deployed engineer works inside the customer's project and writes the code that gets the product running. At the other, Mistral AI's deployment strategists "drive the adoption and deployment of Mistral's AI solutions, working closely with customers from strategic vision to production implementation".

• Suits you if: you are an engineer who likes customers, or a consultant or ops person who likes technology.
• Typical titles: Forward Deployed Engineer (Anthropic, Lovable, Cursor, n8n); AI Deployment Strategist (Mistral AI, Pigment); Solutions Engineer (Dust, Legora).
• Skills postings ask for: Python (59%), then SQL (19%) and RAG (17%).
• Who's hiring, and where: Mistral AI leads by a wide margin, ahead of Celonis, OpenAI and Databricks. 16% of roles are fully remote; Paris and London lead, and Munich matters thanks to Celonis.
• Way in: it leans senior, with 11 junior roles. Celonis's Orbit Program for associate solution consultants, run across the UK, DACH, Benelux and the Nordics, is one of the clearest entry points.

"Forward deployed" now appears in more Tokn job titles (43) than "machine learning engineer" (33). Demand is surging elsewhere too: Indeed data reported by Business Insider counted 5,330 forward deployed engineer postings in April 2026, up from 643 a year earlier, according to dev.ua's summary. Those are US and global figures, not European ones.

Browse AI deployment and operations jobs.

AI product manager and designer

Product managers decide what an AI product should do, and designers shape how people use it. Almost 9 in 10 of the path's 76 jobs have no engineering word in the title.

• Suits you if: you already work in product or design and understand how models behave and fail.
• Typical titles: Senior Product Designer (Dust, Legora, Nabla); Product Manager, Vibe (Mistral AI); AI Product Manager (Celonis).
• Skills postings ask for: few postings list tools, so there is no reliable ranking; Figma comes up most.
• Who's hiring, and where: Lovable, Legora, Mistral AI and Synthesia lead: mostly AI-native scale-ups, not the big labs. 14% of roles are fully remote; London and Paris lead, with Stockholm next (Lovable and Legora).
• Way in: the hardest path for juniors, with just 2 junior roles. Expect to need prior product or design experience.

Browse AI product and design jobs.

AI marketing, growth and content

These teams take AI products to market through product marketing, regional B2B marketing and growth. Its 62 jobs include a hybrid role: the growth engineer.

• Suits you if: you are a marketer who can explain a technical product to technical buyers.
• Typical titles: Product Marketing Manager (Gladia, Nabla, Wayve); Growth Engineer (Granola, Lovable); B2B Marketing Manager, DACH (OpenAI).
• Skills postings ask for: too few postings list tools to rank them.
• Who's hiring, and where: Lovable, Mistral AI and Synthesia lead. 10 jobs (16%) are fully remote; Paris and London lead, with Stockholm next.
• Way in: it leans senior, but internships exist, such as Akur8's marketing internship in Paris.

Browse AI marketing jobs.

Who's hiring for AI jobs in Europe

AI-native companies post about 3 in 4 jobs on Tokn, and hiring is concentrated: the top five companies hold 260 jobs, a third of the board.

The list mixes model labs such as Mistral AI and Anthropic, AI-native scale-ups such as Lovable, ElevenLabs and Legora, established software companies such as Celonis, and consultancies such as Artefact. The scale-ups post most of the product, design and marketing roles.

Mistral AI alone holds about 11% of the board, though some of its postings are one role listed in several cities. Work policies differ sharply by employer: ElevenLabs, Cohere and n8n are remote-first, and half of Synthesia's roles are fully remote, while OpenAI, Anthropic and Mistral AI hire almost entirely on-site or hybrid.

For contrast, Glassdoor's 2018 article on AI jobs in the UK counted 120 AI jobs in the whole country, with Babylon Health as the top employer. Tokn, tracking only 86 companies, lists 224 in London alone. The methods differ, but the direction is clear.

Where the AI jobs are: London, Paris and remote

London (224 jobs) and Paris (221) are neck and neck, and together hold more than half of the board. About one job in six (132) is fully remote. Next come Stockholm (43), Munich (26), Madrid (25), Amsterdam (14), Dublin (13) and Berlin (11).

Two caveats on remote. About 40% of postings don't state a policy; among those that do, 28% are fully remote. And "fully remote" often means within listed countries or EMEA time zones, so check eligibility. Smaller city counts mostly reflect which companies we track: Berlin's 11 jobs say more about our coverage than about its AI market.

How to choose your AI career path

Start from what you already do well.

Then check seniority. About 69% of roles with a stated level are senior or lead, and the typical minimum, where given, is around 5 years. Junior roles are about 9% of all jobs and internships about 5%. If you are starting out, we think your best bets are machine learning, data science, internships and associate programmes, because that is where the entry-level openings are.

For skills, learn Python first: it appears in 58% to 75% of postings in every technical path. Then specialise: PyTorch for research, Kubernetes and Terraform for infrastructure, TypeScript for LLM product work, SQL for data and operations.

Find AI jobs and get alerts on Tokn

Each path above links to its live listings. Pick the one or two that fit you, then create a job alert so new roles reach your inbox.

FAQ

Can I get a job in AI without coding?

Yes. Product, design, marketing and deployment strategist roles exist, and about 1 in 5 AI jobs on Tokn is a business-facing role with no engineering in the title. Many still expect technical fluency, and much of the deployment and operations path is engineering work.

Do I need a PhD to work in AI?

Only for some research scientist roles. On Tokn, research engineer titles outnumber research scientist titles more than 3 to 1, and most other paths ask for experience rather than degrees.

What is the best entry-level AI job?

Data science and machine learning have the largest shares of junior roles, and machine learning has the most internships (11). Internships make up about 5% of all postings on Tokn. Associate programmes, such as Celonis's Orbit Program, are another way in.

Can I work in AI fully remotely from Europe?

Sometimes, but it is the exception: about one job in six on Tokn is fully remote, and hybrid is more common. LLM & GenAI is the most remote-friendly path (26%) and data science the least (10%). Remote-first employers such as ElevenLabs, Cohere and n8n are your best bet.

Is AI a good career in 2026?

Globally, demand looks strong. PwC's 2026 Global AI Jobs Barometer, based on over 1 billion job ads in 27 countries and territories, found that jobs requiring AI skills grew 69%, against 9% for the overall market, with a 62% wage premium for AI skills. Those are global figures with no European breakdown, and few European AI postings publish pay, so check salaries role by role.