Two vocabularies, and only one of them is real
Every company has a title it prints on your email signature and a level it uses internally for pay, promotion and headcount. Titles are marketing. Levels are the machinery.
Large technology companies run numbered ladders: L3, L4, L5, or SDE I, II, III, or IC1 through IC6. Banks and insurers run grades. Consultancies run analyst, consultant, manager, principal. Start-ups often run nothing at all and hand out titles at hiring time because a title is free and salary is not.
The consequence is that a Senior Data Scientist at a thirty-person company and a Data Analyst II at a bank may be doing identical work, and the bank’s analyst may have more responsibility. Neither title tells you anything until you know the scope behind it.
What actually separates levels
Not years. Not tools. Not how many algorithms you can name. Levels are defined by how much ambiguity you absorb and how far the consequences of your work travel.
- Junior. Given a well-defined task, you produce a correct result and say when you are stuck. Somebody else decided what to do and reviews what you did.
- Mid. Given a problem, you choose the approach, do it, and finish it without supervision. You are trusted to notice when the approach is wrong halfway through.
- Senior. Given a vague area, you define the problem, get other people to agree it is the right problem, and deliver something that keeps working after you stop looking at it. Much of your output is other people’s changed behaviour.
- Staff or principal. You change what the organisation works on. You are measured on decisions avoided as much as work done.
Notice that only the first level is about producing correct code or correct numbers. Everything above it is about deciding what should be produced. That is why people plateau at mid-level with excellent technical skills, and why the fix is never another framework.
The manager fork is a fork, not a step
At roughly senior level most organisations offer two paths: keep building, or manage people. They are not up and further up. They are different jobs with different daily content (one-to-ones, hiring, performance conversations, planning) and the skill overlap is smaller than people expect.
Two things worth knowing before you take the fork. First, many companies pay the individual-contributor track and the management track equivalently for several levels, so managing is not a raise. Second, returning is possible but costs you time: a manager who stops writing code for three years and goes back typically re-enters a level below where they left.
Title inflation, and what it costs later
Small companies inflate titles because a title has no cash cost. A twenty-person start-up will happily call a person with three years of experience Head of Data.
This is fine right up until you apply somewhere else. Two things then happen. A large employer maps you to its own ladder using scope, not your title, and frequently offers you a level below what your title implied: this is called down-levelling, and it is routine, not an insult. Worse, an inflated title on a CV invites interviews for jobs that expect the scope the title claims, and those interviews end badly for everybody.
The same effect runs across markets. Moving from a services firm to a product company, or from one country to another, usually costs a level, because scope in the new place is measured differently. Expect it and it stops feeling personal.
Two practical uses
When reading an advert, translate the responsibilities into the ambiguity ladder above and ignore the title. “Partner with stakeholders to define measurement strategy” is senior scope whatever the header says. “Build reports according to specification” is junior scope even if the header says Senior Analyst. Apply to the scope you can actually hold, plus one step.
When negotiating, negotiate the level first. Salary bands are attached to levels, and a recruiter usually cannot move you far inside a band but can sometimes place you at the next one. Asking “which level is this role, and what distinguishes it from the one above?” is a normal question, it signals that you understand how the company works, and the answer tells you your realistic ceiling before you spend two weeks on interviews.
Every lesson here is open. Your progress is recorded once you pass the AI at Work exam.
Check your understanding
Daniel has been Head of Data at a twenty-five person start-up for two years, managing nobody and building dashboards and pipelines himself. A 4,000-person company interviews him and offers a role at their mid-level engineer grade. What is the most accurate reading?
Pick the one you would defend. Your answer stays private.
Stuck on something?Ask Rukmini, your AI tutor