AI Application Developer
Build real software on top of LLM APIs — chatbots, RAG assistants, tool-using agents. The most in-demand technical entry into AI right now, and one where a strong portfolio often beats a formal ML background.
Typical pay: €55k–€85k (IE) · £45k–£75k (UK) estimate — verify before relying on it
The short version
Right now, AI Application Developer is the most in-demand technical way into AI work. You're not training models from scratch — you're building useful software on top of the models that already exist. And in this field, a portfolio of working apps often opens more doors than a formal machine-learning background.
A day in the life
You're building a RAG assistant that answers questions over a company's documents. You wire up tool calling so it can actually do things, not just talk. You run evals to catch where it goes wrong and tune the cost so it's affordable to run. You ship, watch how people use it, and iterate. It's software engineering — with AI as the most powerful new building block.
Who it suits
You'll do well if you like building, can learn to code (or already can), and would rather ship than theorise. If you've never coded, that's your Foundations stage — it's real work, but it's a well-worn path, and this is the destination many people are aiming for.
Is this you?
Find your path → to check it's your best fit, or talk to a mentor → to map the route from where you are.
An honest note. IE/UK salary and demand figures are realistic estimates, not guarantees — and if you're starting from zero coding, be realistic that the foundations take real time. A strong portfolio is what gets you hired.
Your roadmap
Foundations
Get the programming basics in place.
- Learn Python or JavaScript fundamentals
- Understand APIs, HTTP and version control with git
- Get comfortable reading docs and debugging
Core skills
Learn to build with LLMs.
- Work with LLM APIs (OpenAI, Anthropic)
- Prompt engineering in code, not just a chat box
- Build RAG — embeddings and a vector database
- Use function/tool calling and basic agents
- Learn evaluation and cost control
Portfolio
Ship things people can actually use.
- Build 2–3 real apps (a RAG assistant, a tool-using agent)
- Deploy them publicly with clean code on GitHub
Job-ready
Target AI engineering roles.
- Lead with GitHub and live demos
- Target "AI engineer", "LLM developer" and "GenAI" roles
- Remember — many teams take a strong portfolio over a formal ML degree
What you'll need to start
- Some programming (Python or JavaScript) — or willingness to learn it first
- Logical, problem-solving mindset
- Comfort with building and debugging
Related paths
AI Ops / MLOps
Deploy, monitor and keep AI systems reliable in production. If you've got an infrastructure or DevOps background, this is one of the fastest-growing and most natural ways to move into AI.
ML / Data Practitioner
The deeper data-science and machine-learning route — the longest path here, with the highest ceiling. For people who like maths, data and patience, and want to build models, not just use them.