Data Engineer / Applied AI
Pipelines, warehouses, models, agentic apps. Hands-on portfolios outperform multi-cert stacks here.
The genuinely interesting lane, but the AI bit is increasingly product work that needs real engineering chops. Drift toward data engineering if you want stable hiring.
You like working close to the data, you can write Python and SQL fluently, and you'd rather ship a working pipeline than train a model from scratch. Data engineering is where the actual jobs are.
You want to do research-style ML. Those roles exist, they're rare, and they hire from PhD pipelines rather than career-changers.
Phased progression
Foundations → first role → specialisation → advanced. The realistic order, not a script.
- 010–6 monthsFoundations
Literacy, lab habits, the cert that opens first conversations.
DP-203 - 026–18 monthsFirst paid role
Land a Data Analyst → Data Engineer. Operational time, not more certs, earns the next move.
Data Analyst → Data Engineer£35–50k analyst - 031.5–3 yearsSpecialisation
Add a specialist credential aligned to the work you're already doing.
AI-900£55–80k data eng - 043+ yearsAdvanced
Move into adjacent roles. Long-term credentials become worth their cost.
Analytics Engineer£90–130k senior / ML eng (UK)
- 01Foundations0–6 months
Literacy, lab habits, the cert that opens first conversations.
DP-203 - 02First paid role6–18 months
Land a Data Analyst → Data Engineer. Operational time, not more certs, earns the next move.
Data Analyst → Data Engineer£35–50k analyst - 03Specialisation1.5–3 years
Add a specialist credential aligned to the work you're already doing.
AI-900£55–80k data eng - 04Advanced3+ years
Move into adjacent roles. Long-term credentials become worth their cost.
Analytics Engineer£90–130k senior / ML eng (UK)
Certification sequence
Ordered by realistic relevance, not vendor marketing.
- DP-203
- AI-900
- Databricks Data Engineer Associate
- Snowflake SnowPro Core
- AWS Data Engineer Associate
Practical projects
What to actually build, the portfolio that opens interviews.
- Build a small dbt project on Snowflake or DuckDB
- Author an end-to-end pipeline (Airflow or Dagster) with tests
- Ship a RAG demo with eval + observability, not just a notebook
- ·Analytics-engineering-first via SQL
- ·ML via research background
Realistic expectations
What no recruiter will tell you.
That a stack of AI certificates substitutes for shipping something. It doesn't. A public repo with a real pipeline, a real model and a real evaluation beats every credential in this lane.
Twelve to twenty-four months from a strong analyst or backend role. From a standing start it's longer, and the market for true entry-level data roles is thinner than the LinkedIn noise suggests.
The next step
The pathway is plausible. Whether it holds for five years is a different question.
A Career Verdict applies the framework to your actual stage and stack: what holds, what breaks, what would change the call.
A route shows what people usually do. A Career Verdict judges whether it's realistic for you.
A Career Verdict includes
Built on POST's practitioner-authored assessment framework, calibrated by James from twenty years across helpdesk, infrastructure and security. Framework is human-authored; the verdict applies it to your inputs.