Data Engineer

1What you'd actually do

• Writing and editing data pipelines - code (usually Python and SQL) that extracts data from a source system, reshapes it and loads it into a warehouse on a schedule. A lot of this is modifying pipelines that already exist rather than building new ones from scratch.
• Investigating why a pipeline failed overnight: a source system changed a column name, an API stopped responding, a file arrived late or malformed. Working out what broke, fixing it, and backfilling the missing data.
• Writing SQL transformations that turn raw messy tables into clean, well-named tables that analysts can use without needing to understand the underlying system.
• Talking to the people who will use the data - an analyst who needs a new field, a finance team whose monthly report doesn't reconcile - to work out what they actually need before building anything.
• Code review: reading colleagues' pull requests and having your own reviewed, since most teams work in Git with a proper review process.
• Adding tests and monitoring - checks that row counts look sane, that key fields aren't suddenly null, that a daily job actually ran - and responding when alerts fire.
• Working with cloud infrastructure: setting up storage, compute, scheduling and permissions, often through configuration files rather than clicking around in a console.

2How graduates get in

• Graduate schemes in technology, data or engineering at large employers - banks, insurers, retailers, telecoms, media, consultancies and the larger tech companies. Some recruit specifically for data engineering; more commonly you join a broader tech or data intake and are streamed into data engineering after an initial rotation. This is a common route.
• Direct entry as a junior or graduate data engineer at smaller companies, scale-ups and data consultancies. These roles are advertised year-round rather than in an autumn recruitment cycle, and usually involve a technical test rather than a full assessment centre. Very common, and often the faster way in if you already have coding skills.
• Sideways from an analyst role. Many data engineers start as a data analyst or BI (business intelligence) analyst, get good at SQL, start automating their own reporting, and move across within a year or two. This is one of the most frequent routes and is worth knowing about if you can't land an engineering role directly.
• Sideways from software engineering - if you can already write production code, the gap to data engineering is mostly about learning SQL, warehousing and pipeline tooling. Reasonably common.
• Degree apprenticeships and higher apprenticeships in digital and technology solutions, some with a data specialism. These are an alternative to the graduate route rather than an addition to it, but graduates occasionally enter through similar structured conversion programmes.
• Conversion routes: an MSc in data science, data engineering or computer science, or an intensive bootcamp. These can work, particularly if paired with a portfolio of real projects, but a bootcamp alone is a weaker signal than demonstrable code. Less reliable than the routes above.

3What employers ask for

• A degree in a numerate or technical subject is the usual expectation - computer science, software engineering, maths, physics, engineering, statistics or similar. The subject matters more here than in many graduate jobs, because employers want evidence you can code. That said, plenty of data engineers come from other backgrounds having taught themselves; you'll need a portfolio to compensate.
• Typically a 2:1, though many employers - especially smaller ones and those using technical tests - care far more about whether you pass the coding assessment than about your classification. Some larger schemes still screen on grades including UCAS points or A-level results.
• Working SQL. This is close to non-negotiable. Employers expect you to write joins, aggregations and window functions without looking everything up.
• At least one programming language used properly - Python overwhelmingly, sometimes Scala or Java in older or high-volume environments.
• Evidence you've built something: a personal project that pulls data from an API and stores it somewhere, a dissertation with real data wrangling, an internship, a placement year, contributions to open-source. Employers vary in how heavily they weight this, but for direct-entry roles it often matters more than anything on your transcript.
• Cloud familiarity (one of the major cloud platforms) is frequently asked for but is often treated as 'nice to have' at graduate level - most employers accept they'll teach you their stack. Professional cloud certifications help more with smaller employers and consultancies than with large graduate schemes.

4Skills that matter