HN user
Yeah that's fair -- maybe just focusing on becoming an expert in data and ML engineering in the long-run is a more prudent career choice than totally dropping those for another corner of engineering. There's some wisdom here.
Taking that much time off of any job is pretty hard to do!
The real goal of a bootcamp though (besides networking) is adding a project to your portfolio. You can then show off this project to hiring managers to demonstrate your competence.
You don't need to do a full-time bootcamp to make this happen! I just checked and Flatiron school and Hack Reactor both have online part-time full-stack bootcamps. Also, while not as big of an investment, online platforms like Codecademy have career paths with a capstone project you can complete and add to your portfolio.
So, overall there are a variety of pathways to do this. Totally fair to point out though a bootcamp of some kind is a nice option in the off-chance it's a possibility.
You're right, Google has been very supportive -- very thankful for that! That said, if you work at a technology company, you CAN build a portfolio and self-advocate for projects with engineering teams your data science team already works with.
I think it's worth seeing how far you can get with this approach before interviewing for engineering roles at other companies.
One notable caveat with my approach: pursuing a portfolio approach took me a whole year to switch from DS to engineering. As you point out, grinding leetcode may only require half that time.
Yep. In any case, building a model is a just part of the puzzle -- serving it in production is a whole other skillset that is very valuable. imho
Yes -- one thing I struggle with is choosing whether to work on data engineering projects (which I know I'm good at and can provide immediate value) or challenging myself with more "traditional" full-stack engineering work where I can learn a lot.
Maybe a 60/40 mix is optimal
Oh, I'm sorry about that! Here's a friend link that should allow you to bypass any paywall (if you've already read a couple paywalled posts): https://zachary-thomas.medium.com/how-i-became-a-software-en...
Here's an earlier draft of the post on Quip: https://gstudent.quip.com/ILF1ABpvjsSh/Can-a-Data-Scientist-...
Hey everyone, OP here -- thanks for reading or discussing.
Curious if anyone else has switched from data science to software engineering, or has thoughts on comparing the two as a career choice. I'm all ears!
Sure! My company calls the role "software engineer" so that's how I describe the role but I think developer is also a good word for it
Hi! I'm Zach, the author -- I put together this doc since I couldn't find many resources I could recommend on metric design and evaluation, which pops up in data scientist/product analyst interviews and is a common task for data science teams.
Looking to hearing your feedback and if you've seen other good resources on this topic!
Hi! I'm Zach, the author -- I put together this doc since I couldn't find many resources I could recommend on metric design and evaluation, which pops up in data scientist/product analyst interviews and is a common task for data science teams.
Looking to hearing your feedback and if you've seen other good resources on this topic!
Awesome! Some quick reactions:
- Wish I could upload a photo instead of taking one on the spot!
- Took me a little bit of time to get past the "what city are you in" screen (the selector was at the bottom of the page, didn't see it
- It will be cool when the memes I really liked can be easily shared
Looking forward to seeing this develop!
Thanks, anonfunction! I've corrected it
Agreed! Pandas is great but at this point instead of using logic to solve data structuring tasks I often find myself googling for an optimized built-in Pandas method to help me out. Leads to less elegant code -- not sure if it is less readable though.
Sure! Practically speaking, as a data analyst, I would probably notice a missing month when plotting the trend of MAU over time.
(You can make "but what if a month is missing?" a latter part of a multi-part interview question)
Generally I would assume that data engineers would have a month of no users set to zero or that I could ask them why that's not the case and note that for future reference.
Many data analysts use exclusively SQL and R/Python
One thing I mention in the doc intro is that SQL questions are one part of the data analyst interview, not all of it. If this doc helps people get through the SQL questions so that they can spend more time being thoughtful with open-ended analytics questions, that's good thing (both for interviews and for work more generally)!
Yes, in particular for data analysts using SQL to answer product/business questions in a one-off fashion
Sorry about that -- in real life you would just ask the interviewer what MAU meant if they didn't spell it out.
The purpose was to make the questions more realistic, since at least in my experience in data analyst interviews the questions are asked in the context of actual business or product situations ... like company leaders, PMs, or others wanting to understand trends in MAU.
To your point, I personally write SQL for analytics and product/business purposes, not system monitoring
Sure, that's a fair criticism. That said, you can build multi-step interview problems with SQL (I tried to convey one or two in this doc) such that interviewers can build up towards needing a more advanced window function instead of starting there.
I've used BETWEEN ROW maybe once or twice in my career in a professional setting. Self-joins more often, but as others have pointed out window functions are more efficient here for writing dashboard ETLs, etc.
Btw, are you minimaxir who wrote gpt-2-simple? I was looking at your tutorial a month ago while putting together a solution for the Kaggle COVID-19 NLP challenge!
Thanks for the feedback! The first section is intentionally about self-joins since they get asked about in interviews, but other people have brought up that window functions are more efficient, so I'll add in those solutions as well.
Sorry you feel that way! Thankfully my employer felt differently :)
Good q! +1 to the other replies on this comment. I have two points worth mentioning:
(1) The flavor of SQL I use at work supports macros, which are functions that can take in parameters like a function in R/Python might. So, the SQL is "turbo-charged" in that sense and some of the value-added of switching over to Python/R is diminished.
(Big Query has UDFs, which seem similar: https://cloud.google.com/bigquery/docs/reference/standard-sq...)
(2) Like I mentioned in the doc, I personally use the SQL in these practice problems for ETLs on dashboards showing trends. AFAIK, much easier/efficient to write metrics for daily ETLs in SQL than R or Python, especially if these are top-line metrics like MAU.
Hmm, haven't seen those a ton at work / in interviews! Would love to see some examples though if you have any.
Mode's SQL tutorial uses SUM(CASE ...) and CROSS JOIN to mimic pivots: https://mode.com/sql-tutorial/sql-pivot-table/
Hey! MAU = Monthly Active Users. Sorry about that, I think this could be described as data analyst/analytics jargon. Good feedback to spell these out, thanks!
Yep, Ollien made a similar point. Currently some of the questions have expected output but not all.
Yes -- making this a proper webpage with an embedded SQL editor like selectstarsql.com does would be a good next step
Good catch -- I can add that in as an alternate solution. Thanks!
Yeah, it's because (1) I started this doc over a year ago when I was still at Salesforce (which owns/employees use Quip) and (2) AFAIK Google Docs still doesn't have native code block support :/
Some of them have sample output but yes, it would be helpful to extend this to all examples. Good feedback!
Great resource! I have personally found the top Leetcode SQL questions to be hit-or-miss without curation (too easy, too hard, doesn't cover topics like window functions / self-joins like I would encounter in interviews, etc.) but you're right, it is a great, interactive interview resource