Every application starts with a job description. Most people read it once, quickly, and ask one question: “Do I have these skills?” That question comes too early. A job description is a message from a team that has a problem. First find the problem. Then you can see which of your experience matters, which lines are important, and which lines are only there because they are in every posting.
Think of a menu in a new restaurant. It lists 40 dishes, but the kitchen is really good at five of them, and the waiter can tell you which five if you ask. The menu also does not say how long you will wait, or that the kitchen closes early on Mondays. A Data Scientist posting is often like that menu: long, a little proud of itself, and silent about some things that matter. In this guide, we read it carefully, we notice what it leaves out, and we decide what to ask.
The job description below is for a company that does not exist. We built it from 15 real Data Scientist postings at fintech companies, product companies, an insurer, a bank and a language-learning app. We kept their real patterns: the clear parts, the vague parts, and the small problems that real postings have. Then we put it in one company, so that the whole guide can use the same team and the same product.
Parlo is a mobile app for learning languages. About 3 million people use it every month, in 60 countries. Most of them learn English, Spanish or French. The app is free to start, and people can pay for Parlo Plus, a subscription with more features. Like most subscription apps, Parlo depends on two numbers: how many new learners stay, and how many of them start to pay.
Job description
Data Scientist, Growth & Learning
Parlo · Toronto, Ontario
Remote in Canada, with 2 team days per quarter in Toronto
Parlo helps 3 million learners a month in 60 countries speak a new language with confidence, a few minutes a day. We are a data-driven, AI-first company, and every product decision starts with evidence. Our Growth & Learning data team works with product, design, engineering and our learning-design team to help learners stay motivated and make real progress.
As a Data Scientist on Growth & Learning, you will be the DRI for data in your product area. You will own analyses end to end, from the first question to the decision, and you will help shape how Parlo measures success for new AI-powered features, such as our new AI conversation practice.
This posting is for an existing vacancy. Canadian work experience is not required. We use AI tools to help sort applications; a person reads every application before any decision. If you need an accommodation at any stage, tell our recruiter.
Composite job description for a fictional company, built from 15 real Data Scientist postings.
Before you compare the posting with your CV, answer three questions about it. Write your answers down. It takes about ten minutes, and then you can compare them with ours below.
Which decisions will this person help to make? Use one sentence, in plain words, with no tool names.
Who uses the work, and what happens when it is wrong?
Which three lines would you remove, if you had to keep the posting honest and short?
Our answers
1. The decisions. Parlo changes its app all the time: a new lesson format, a new onboarding screen, a new price, a new AI feature. For each change, somebody must decide: keep it, change it, give it to everyone, or remove it. This person helps the product managers make those decisions with evidence, and says clearly how sure the evidence is. Almost every line in “What you will do” is a way to do this.
2. The users, and the cost of a mistake. The first users of the work are product managers. They usually do not read SQL or confidence intervals. They read your recommendation. Behind them are 3 million learners. If you say that a feature works when it does not, Parlo gives it to everyone, and many learners leave without anyone noticing why. If you say that it does not work when it does, the team throws away months of good work. There is also a cost in money: the posting mentions AI conversation practice, and AI features cost money every time someone uses them. So a wrong analysis can be expensive in two ways.
3. Lines to remove. “A growth mindset and a passion for languages” tells you about the company’s view of itself, but nobody can check it in an interview. “Data-driven, AI-first” appears in many postings and says little about this team. And the tool list (Snowflake, BigQuery, dbt, Airflow, Spark, Looker, Tableau, Mode, Hex, Amplitude) is almost certainly longer than what the team uses. A team of eight people rarely uses two warehouses and four tools for reports. We come back to this on the next page.
Did you notice the line about the AI conversation practice? A posting often names the project that made the team open the role. That is a good guess about what the interviews will discuss, and in this guide it is exactly what happens.
If your answers are different from ours, that is fine. The point is the habit: start from the decisions and the people, not from the list of tools. In the interviews, the conversation will be about decisions. The tools come up only as ways to reach them.
On the next page, we read the posting line by line and say honestly what each part means: which kind of Data Scientist job this is, what is required, what is only a wish, what is missing, who should apply, and which questions to ask the team. After that, we look at the five interviews in Parlo’s process, and then follow five different candidates through the product case.
