Chris Hua

“Data” “Scientist”

Choosing between data science offers


Not all data science roles are created equal. Everyone will tell you that “data scientists” as a role play varying parts in different companies, but how can you choose between different teams?

Some heuristics I found useful:
* % of employees that are data
* are people promoted from within
* investments in data infra (eg how long does it take to get experiment results)
* data culture - who does data report to in management?
* % of reactive vs proactive work

— Chris Hua (@hingeloss) December 27, 2017

Hilary Mason had a tweet about this that got a lot of replies along the lines of “well I work at this place and it’s obviously a great place for junior data scientists, I’m here!” - but very few actually substantiate what characteristics make them a good fit for junior data scientists.

Some quick background: I graduated this spring from Wharton, after interning with Airbnb Data Science. I had 4 or 5 full-time offers of varying roles from companies of varying sizes and industries, and had to make difficult choices. These are the heuristics that I used; experiences may vary. I’m very happy with my decision to join Quora Data Science and think the team does well on many of these fronts.

These are some recommendations in evaluating what makes a good data science team. These are not really to be considered general career advice - joining an amazing team on a sinking ship isn’t smart, and taking a huge paycut isn’t necessarily healthy either. I also don’t want to create some Manichean divide between ‘junior’ and ‘senior’ roles, hence the quotes.

All levels

All data scientists deserve work that is meaningful, impactful, and valued within the company.

‘Junior’ roles

Junior data scientists should optimize for learning. I’ve written about this before with respect to course selection, but I think that learning is just as important when picking a job.

You can maximize your learning by working somewhere which:

‘Senior’ roles

Senior data scientists should optimize for impact - often by making other people more effective. I don’t think I’m here yet, but there are a few common characteristics of the senior data scientists or data science managers in teams that I appreciate.

Teams where senior data scientists/managers are empowered to drive impact in these ways are well-positioned for success.


Was this useful? Totally off base? Something in between? Let me know - as always, these are strong opinions, weakly held.

  1. e.g. at Airbnb or Pinterest. Quora’s platform is philosophically similar to Airbnb’s.
  2. e.g. Knowledge Repo.
  3. Sometimes this is inevitable. Certain features really cannot be effectively tested, e.g. visible pricing changes. Alternatively, companies which cater towards enterprises will often find it difficult to test large features with any semblance of power, when the correct unit of experimentation is the enterprise.
  4. You know who you are.