Six roles, six entry points
The AI job market in 2026 is a mess of overlapping titles. We grouped them into six distinct roles. The differences are real — different skills, different daily work, different entry points.
1. Applied GenAI Engineer (€60–150k EU)
The fastest-growing AI role of 2025–26. Hybrid: software engineering + prompt engineering + product thinking. Most B2B SaaS companies are hiring at least one. Best entry point if you're already a working software engineer — you can make the jump in ~3 months.
What you'll do daily: wire LLMs into product features, build RAG over your company's docs, design eval sets, ship JSON-mode integrations. Maybe 30% prompts, 50% engineering, 20% product/design discussions.
Path: Prompt Engineering → RAG → Deployment. About 30 hours of structured learning + a side project that ships.
2. ML Engineer (€65–160k EU)
Bridges data science and software engineering. Owns the model in production. Trains, deploys, monitors. The traditional "engineering ML" role.
What you'll do daily: build training pipelines, manage feature stores, deploy models, debug accuracy regressions in prod. Heavier on infra than the GenAI role.
Path: AI Foundations → LLMs & Transformers → Fine-tuning → Deployment & MLOps. ~45 hours.
3. MLOps Engineer (€70–170k EU)
The SRE for ML. Builds the platform other people deploy on. Owns autoscaling, observability, cost.
What you'll do daily: Kubernetes, vLLM/TGI configs, GPU scheduling, cost dashboards, incident response when accuracy drops at 3am.
Path: AI Foundations + Deployment & MLOps + RAG. Strongest pivot from existing SRE/DevOps backgrounds.
4. AI Researcher (€80–250k EU)
Frontier work. Real publications, real ablations. The salary bands are wide because the jobs aren't fungible — the labs all want different specialties.
Honest scope: most positions require a strong publication record or PhD pipeline. We can teach the math and reproduction discipline. We can't substitute for a 6-month original-work portfolio.
Path: AI Foundations (deeply) → LLMs & Transformers → Fine-tuning. Then a research sprint outside any course.
5. AI Security Specialist (€75–180k EU)
Red-teams and hardens AI systems. Prompt injection, model extraction, output filtering, supply-chain. Small role today, growing fast — under-supplied, regulator-favored.
Path: Prompt Engineering (security lessons especially) + LLMs & Transformers + Deployment. Strong pivot from existing AppSec.
6. AI Product Manager (€65–150k EU)
Scopes AI features. Defines quality thresholds. Decides what ships and what gets killed. Most AI orgs need one and most don't have a great one.
Path: Prompt Engineering + RAG + LLMs & Transformers (lighter touch). The technical depth is "build a small AI demo end-to-end" not "fine-tune a model".
The newer roles
We added four more in our latest catalog: Data Engineer (AI-focused), Computer Vision Engineer, AI Solutions Architect, and we still consider Prompt Engineer a real standalone role at junior levels. The full list and roadmap for each is at [/careers](https://nextgenailearn.com/careers).
What hiring managers actually weigh
Beyond the titles, the consistent signal across hiring conversations:
- Shipped something. A blog post + repo + working demo beats a CV bullet. Always.
- Eval discipline. "How did you know your prompt was good?" The answer "I tried it and it worked" gets you rejected. The answer "I built a 50-case eval set, scored on
exact-match, and tracked regressions per case" gets you the next round. - A point of view. Strong opinions, loosely held, on which approach (prompt / RAG / fine-tune) to use when.
- Cost-awareness. Knowing roughly what an LLM call costs and when to tier down. This signals seniority.
Where to start
Pick a role on [/careers](https://nextgenailearn.com/careers). The roadmap card maps it to specific paths. The first lesson of the first recommended path takes 8 minutes. By the end of the first week you'll know whether the role fits.