Expect to write SQL in front of someone
Every data analyst loop in this market includes live SQL. Not a quiz about syntax — a real question against a schema you are shown five minutes earlier. Practise typing while explaining, because that is the actual skill being tested.
Here are two tables, orders and customers. Give me monthly revenue per city, and only for cities with more than fifty orders.
Narrate as you go. Say what you are joining on and why, and check your own assumption about whether the join can duplicate rows. Interviewers here consistently say the silent candidate is the hardest to assess.
Window functions and joins that do not fan out
These two topics separate the analysts who have worked with real data from those who have worked with tutorials.
- ROW_NUMBER, RANK and DENSE_RANK — and when the difference bites
- Running totals and period-over-period without a self-join
- Why a one-to-many join inflates your SUM, and how you noticed
- LEFT JOIN plus a WHERE on the right table, which quietly becomes an INNER JOIN
This report double-counts revenue after we added the payments table. Why?
Explaining the result
This is the round that decides most offers and the one candidates prepare least. You will be asked to present a finding to someone non-technical — often the person who would actually be your stakeholder.
Lead with the answer, not the method. "Revenue fell 12% in Khujand, and it is almost entirely one customer who stopped ordering" is a good opening. The methodology comes second, and only if asked.
- State the number, then the caveat, then what you would do next
- Know the limits of your own data and say them before you are asked
- Never present a chart you cannot explain the axes of
Tools, and how much they matter
Excel is not a weakness in this market — it is the interface most stakeholders actually use. Power BI is the most commonly requested BI tool across the region, with Tableau appearing more in international-facing companies. Python matters more for cleaning than for modelling in most analyst roles here.
What interviewers here are really assessing
- Can you get to a number without being handed a clean table
- Do you check your own work before presenting it
- Will you say "I do not know yet" instead of producing a confident wrong figure
- Can you turn a vague business question into a query
What to focus on at your level
Junior · 0–2 years
Solid SELECT, GROUP BY and JOIN, plus one analysis you did end to end and can walk through.
Mid · 2–5 years
Window functions without hesitation, data you cleaned yourself, and a finding you presented to a non-technical audience.
Senior · 5+ years
Defining the question rather than answering it, knowing which metric is misleading, and having said no to a request that would produce a bad number.
Open Data Analyst roles right now
Read from the live roles on Nebryx when this page loaded — not a list written into the page. If it is empty, nothing matching is open today.
Frequently asked questions
How much Python do I need?
For most analyst roles here, enough to clean and reshape data — pandas at a working level. Heavy modelling belongs to data scientist postings, which are a much smaller share of the market.
Is Excel experience a negative?
No, and treating it as one is a mistake. Being fluent in the tool your stakeholders already use is an advantage, as long as SQL is where your analysis actually happens.
What if I freeze on the live SQL round?
Say what you are trying to do in words. Interviewers routinely pass candidates who got stuck on syntax but described the right approach, and fail candidates who wrote something that ran and was wrong.