Data Minds: Emily Fay — Data Scientist at OpendoorData Minds is a series that profiles professionals working with data.
You’ll have a chance to learn about their story, what a typical day looks like, as well as their advice for others.
Andrei LyskovBlockedUnblockFollowFollowingMar 29Recently raising $300M(3.
8B valuation), Opendoor is one of the leading companies in the online home-selling marketplace space.
Based in San Francisco, the company also offers many interesting data science problems to work on.
Emily Fay has been a data scientist with Opendoor for over 2 years now.
Before making the jump to tech, she went through the Insight Data Science program and completed her PhD at Stanford in geophysics.
In this article we talk about her transition from academia to tech, the value of the Insight program, the importance of creating and reading documentation, as well as general advice she’d offer to others.
Breaking Out of AcademiaAfter completing her undergraduate degree in geological engineering at Queen’s University, Emily went on to do her PhD at Stanford in geophysics.
During her time at Stanford, she ended up doing a lot of coding, from visualizing data to writing scripts to help with data processing.
It turned out that the coding part was the part of her PhD she found herself enjoying the most.
So despite having applied to post-doc roles, she started considering roles in tech that were suited for PhD’s, such as data science.
However, looking at some of the application requirements for a data science role, she felt intimidated by the long list of technical skills.
To help make the leap into tech, she decided to join the Insight data science fellowship, a program geared towards PhD’s who want to rebrand themselves as data scientists.
This turned out to be the perfect choice for her.
With the serious interview prep that Insight offered, Emily was able to get to a point where she was ready to handle technical interviews with confidence.
Having a structured learning environment and opportunities to practice with peers was also a great help.
Insight also offered opportunities to meet and interact with companies in less formal settings.
This made it a lot easier to get candid thoughts on the work and culture of the company.
Most importantly, Insight provided a network of individuals similar to her (PhD’s looking to pivot into tech).
As a result, she was able to make a lot of close friends who had similar career trajectories as herself.
Beyond her own cohort, Insight also has alumni at over 200 companies from Airbnb and Amazon to NBC and Reddit.
Tapping into this network and getting their advice was an invaluable resource.
Thanks to the interview prep and network, Emily was able to interview with a diverse range of companies.
She settled on Opendoor, after using a combination of gut feeling and quantitative analysis, such as ranking her choices by location, culture and the type of work.
Data Science At OpendoorIn her first few weeks, she spent a lot of time reading through the codebase, figuring out how things worked and how they were connected.
She also learned as much as she could about the business and the current projects, reading through shared documents and talking to coworkers across the company.
One technique she used was adding sticky notes to her computer for code snippets, and shortcuts for the command line.
Once she had internalized a particular sticky note, she would replace it with a new one.
This allowed her to get up to speed with tools that enabled her to accelerate her understanding of the codebase.
Opendoor also assigned her a code buddy, which she used extensively to shore up knowledge gaps.
In her first year, she felt uncomfortable giving code reviews.
But after spending enough time reading about coding practices and contributing to the codebase, she felt a lot more confident in asserting her views.
Everyone has their perspective on code review, but for her, it was important to get to a point where she could have different opinions on things and be able to back them up.
Her day to day consists of prototyping and productionizing machine learning models for various business applications, building data pipelines, developing metrics and performing ad-hoc analyses for business stakeholders.
As a result, any given week or month may look quite different from the previous ones given the diverse nature of her work.
The common theme among her work is getting to know the problem and building domain expertise.
The latter happens organically, but making sure to read documentation, internal newsletters, paying attention at all-hands meetings, talking to stakeholders, and monitoring Slack channels are all great ways to build domain expertise.
This also means being proactive in your learning and reaching out to relevant people when you don’t understand something.
Emily is also involved with the data evangelism at her company.
This involves both building internal tooling to help non-technical colleagues to interact with data and also creating comprehensive documentation.
By doing this, she was able to cut down a lot of common questions by referring to written documentation.
A general rule is that if you get asked a variation of the same question three times, it’s a good idea to document it.
Opendoor also does sharing sessions, where data scientists have a chance to show their work and get feedback from colleagues.
Advice for OthersSince the field of data science is so vast, it’s unrealistic to expect to be able to memorize everything.
Thus Emily espouses the importance of knowing how to effectively navigate resources such as StackOverflow and various types of documentation (from programming libraries to company-specific documentation).
She also recommends taking time to deeply understand a problem before starting a project.
Making sure to align the vision and expectations on a project is a great way to avoid problems down the road.
Expectations may also change so it’s a good idea to revisit them throughout the project.
The worst thing you can do is build something that nobody ends up using.
In terms of advice related to dealing with your manager, it’s best to think of them as a resource and not someone who’s checking up on you.
The best managers have a good high-level view of what’s happening in other parts of the company and how your work fits into the bigger picture.
As a result, they’ll be more likely to know who you should talk to in order to continue making progress, in addition to helping make things happen, for example getting access to data.
Another important point is to take charge of your career progression.
This means understanding the expectations of your role, and identifying areas where you can improve.
Companies often have rubrics for quarterly/yearly evaluations so it’s a good idea to know what those are and how you’re progressing.
You should never be surprised when evaluations come around, so make sure to take proactive steps to solicit feedback.
Another important piece of advice is to recognize when machine learning may not be a good solution for a particular problem.
Often times relatively simple logic rules may be all that’s required to address a problem.
Being able to empower humans to make better decisions can trump an over-engineered machine learning algorithm.
Finally, if you’re looking for data-related jobs, it’s good to know what you’re applying for.
Often times, companies may have a data science position that at another company would be called a machine learning engineer.
Thus it’s important for you to take time to read the requirements and talk to the recruiter to avoid any confusion during interviews.
She also mentioned keeping an open mind to roles adjacent to data science.
There are way more data engineering roles, and the skills you learn are often transferable to that of a data scientist.